Lasers in the Conservation of Artworks VIII

LASERS IN THE CONSERVATION OF ARTWORKS VIII

LACONA8.indb i

11/10/2010 5:36:01 PM

PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON LASERS IN THE CONSERVATION OF ARTWORKS VIII (LACONA VIII), 21–25 SEPTEMBER 2009, SIBIU, ROMANIA

Lasers in the Conservation of Artworks VIII

Editors

Roxana Radvan National Institute of Research and Development for Optoelectronics—INOE 2000, Romania

John F. Asmus Physics Department, University of California, San Diego, La Jolla, CA, USA

Marta Castillejo Instituto de Química Física Rocasolano, CSIC, Madrid, Spain

Paraskevi Pouli Institute of Electronic Structure and Lasers, Foundation for Research and Technology—Hellas, Heraklion, Crete, Greece

Austin Nevin Courtauld Institute of Art, London, UK

LACONA8.indb iii

11/10/2010 5:36:02 PM

CRC Press/Balkema is an imprint of the Taylor & Francis Group, an informa business © 2011 Taylor & Francis Group, London, UK Typeset by Vikatan Publishing Solutions (P) Ltd., Chennai, India Printed and bound in Great Britain by Antony Rowe (a CPI Group Company), Chippenham, Wiltshire All rights reserved. No part of this publication or the information contained herein may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, by photocopying, recording or otherwise, without written prior permission from the publisher. Although all care is taken to ensure integrity and the quality of this publication and the information herein, no responsibility is assumed by the publishers nor the author for any damage to the property or persons as a result of operation or use of this publication and/or the information contained herein. Published by: CRC Press/Balkema P.O. Box 447, 2300 AK Leiden, The Netherlands e-mail: [email protected] www.crcpress.com – www.taylorandfrancis.co.uk – www.balkema.nl ISBN: 978-0-415-58073-1 (Hbk) ISBN: 978-0-203-81866-4 (eBook)

LACONA8.indb iv

11/10/2010 5:36:02 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Table of contents

Preface

ix

Permanent scientific committee

xi

Gauguin, Mucha, and Art Nouveau J.F. Asmus

1

Innovative approaches in laser cleaning researches and instrumentation development The effect of ultrafast lasers on laser cleaning: Mechanism and practice K.G. Watkins, P.W. Fitzsimons, M. Sokhan & D. McPhail Spectral analysis of the effects of laser wavelength and pulse duration on tempera paints M. Oujja, M. Castillejo, P. Pouli, C. Fotakis & C. Domingo The role of the substrate in the laser cleaning process: A study on the laser assisted removal of polymeric consolidation materials from various substrates S. Kogou, A. Selimis, P. Pouli, S. Georgiou & C. Fotakis

9 15

23

Compact short pulsed fiber laser offers new possibilities for laser cleaning J. Hildenhagen & K. Dickmann

29

Decontaminating pesticide-exposed museum collections J.F. Asmus

33

Laser cleaning of burial encrustation and aged protective coating on Egyptian leather: Optimization of working conditions A.A. Elnaggar, P. Pouli, A. Nevin, M.A. Fouad & G.A. Mahgoub

39

The practical use of lasers in removing deteriorated Incralac coatings from large bronze monuments A. Dajnowski & A. Lins

47

PROCON TT 49: Laser cleaning of ancient Egyptian wall paintings and painted stone surfaces B. Graue, S. Brinkmann & C. Verbeek

53

The influence of paper type and state of degradation on laser cleaning of artificially soiled paper S. Pentzien, A. Conradi & J. Krüger

59

Laser cleaning studies for the removal of tarnishing from silver and gilt silver threads in silk textiles B. Taarnskov, P. Pouli & J. Bredal-Jørgensen

67

Thickness of ablation control by structured light method R. Sitnik, J. Rutkiewicz & J. Marczak

75

213 nm and 532 nm solid state laser treatment of biogenetical fibrous materials M. Forster, S. Arif, C. Huber, W. Kautek, S. Bushuk, A. Kouzmouk, H. Tatur & S. Batishche

79

Free-running Er:YAG laser cleaning of mural painting specimens treated with linseed oil, “beverone” and Paraloid B72 J. Striova, E. Castellucci, A. Sansonetti, M. Camaiti, M. Matteini, A. deCruz, A. Andreotti & M.P. Colombini

85

v

LACONA8.indb v

11/10/2010 5:36:02 PM

Studies on the UV femtosecond ablation of polymers: Implications for the femtosecond laser cleaning of painted artworks I.A. Paun, A. Selimis, G. Bounos & S. Georgiou

93

Monitoring the laser cleaning process of ornamental granites by means of digital image analysis J. Lamas, A.J. López, A. Ramil, B. Prieto & T. Rivas

99

Optimization of laser cleaning parameters for the removal of biological black crusts in granites A.J. López, J. Lamas, A. Ramil, A. Yáñez, T. Rivas & J. Taboada Bronze putti from Wilanów Palace garden façade—conservation studies and tests of laser cleaning H. Garbacz, E. Fortuna, Ł. Ciupiński, K.J. Kurzydłowski, A. Koss, J. Mróz, A. Zatorska, K. Chmielewski, J. Marczak, M. Strzelec, A. Rycyk & W. Skrzeczanowski

105

111

Comparative studies: Cleaning results of short pulsed Nd:YAG vs. fibre J. Hildenhagen & K. Dickmann

119

Laser cleaning of iron: Surface appearance and re-corrosion of model systems C. Korenberg & A.M. Baldwin

123

Reversion of darkened red lead-containing wall paintings by means of cw-laser irradiation: In situ tests and first application S. Aze, J.-M. Vallet, V. Detalle & O. Grauby Comparative study on the irradiation methods against fungal colonization case study S.A. Abd Abd El Rahim

129 135

Investigation and diagnostics methods Absolute LIBS stratigraphy with Optical Coherence Tomography P. Targowski, E.A. Kwiatkowska, M. Sylwestrzak, J. Marczak, W. Skrzeczanowski, R. Ostrowski, E. Szmit-Naud & M. Iwanicka

143

Database of complex paint spectra decomposed by principal component analysis, for identification of artwork colours Zs. Márton, T. Tóth, É. Galambos & R. Mingesz

149

Study of matrix effect in the analysis of pigments mixtures using laser induced plasma spectroscopy M.P. Mateo, T. Ctvrtnickova, A. Yañez & G. Nicolas

155

Pomerania Laboratory—A solution for the cultural heritage research and conservation A. Iwulska, I. Traczyńska, R. Jendrzejewski, M. Sawczak, G. Śliwiński & A. Kriegseisen

161

THz-Time-Domain Spectroscopy—A new tool for the analysis of artwork M.J. Panzner, U. Klotzbach, E. Beyer, G. Torosyan, A. Schmid & W. Köhler

167

19th century paints of Richard Ainè used by Jan Matejko (1838–1983). Analysis of preserved paints from tubes, palettes and of paintings’ surfaces and paint-layer M. Wachowiak

173

Study of the effect of relative humidity on the identification conditions of paper soiling by means of the NIR technique M. Sawczak, G. Rabczuk, A. Kamińska & G. Śliwiński

177

Monitoring, imaging and documentation of artwork Experimentation of a three-focal photogrammetric survey system as non invasive technique for analysis and monitoring of painting surfaces decay condition P. Salonia, A. Marcolongo & S. Scolastico RGB-ITR: An amplitude-modulated 3D colour laser scanner for cultural heritage applications R. Ricci, L. De Dominicis, M.F. De Collibus, G. Fornetti, M. Guarneri, M. Nuvoli & M. Francucci

185 191

vi

LACONA8.indb vi

11/10/2010 5:36:02 PM

3D laser reconstructions of Buddhist temple from Ladakh D. Ene & R. Rădvan

199

Robotized structured light system for automated 3D documenting of cultural heritage R. Sitnik, M. Karaszewski, W. Załuski & P. Bolewicki

203

Through-glass structural examination of Hinterglasmalerei by Optical Coherence Tomography M. Iwanicka, L. Tymińska-Widmer, B.J. Rouba, E.A. Kwiatkowska, M. Sylwestrzak & P. Targowski

209

Editing protocol for the digital mapping of related imagistic investigations L.M. Angheluta

215

U-ITR: A 3D laser scanner prototype aimed at underwater archaeology applications R. Ricci, L. De Dominicis, M.F. De Collibus, G. Fornetti, M. Guarneri, M. Nuvoli & M. Francucci

221

Author index

227

vii

LACONA8.indb vii

11/10/2010 5:36:02 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Preface

The meeting of cultures should offer our children and grand children ‘the privilege of belonging to several worlds in a single life’, to quote a French anthropologist (Serge GRUZINSKI). The International Conference on Laser in Conservation of Artworks—LACONA VIII follows the prestigious series of conferences initiated by Prof. Costas Fotakis organizing LACONA I—1995 in Heraklion, Greece. This was followed by LACONA II—1997 in Liverpool UK, LACONA III—1999 in Florence, Italy, LACONA IV—2001 in Paris, France, LACONA V—2003 in Osnabruck, Germany, LACONA VI—2005 in Vienna, Austria, LACONA VII—2007 in Madrid, Spain. The success of these unique conferences motivated the members of the Permanent Scientific Committee to stress the international addressability of the biennial event and honoured Romanian partners to organize the eigth edition for the first time in one of the East-European countries. The LACONA VIII event—an invitation to dialogue between specialists from different geo-cultural area, and a dialogue between different professional fields for the benefit of a common heritage- was organized in Sibiu, between 21 and 25 September 2009, by INOE – The National Institute of Research and Development for Optoelectronics with precious collaboration of local authorities and cultural institutions. Today, a new affirmation regarding the role of science in Cultural Heritage conservation is almost a truism, but I am sure that the research field dymanics and tendencies are of great interest. I am glad to see that new knowledge boundaries are defined and the papers of scientific community proove a strong and interrelated international activity. The public discussions and the open dialogue during Round Table Session reaffirmed the inexhaustible efforts for correlation of analytical investigation, for trainings and know-how transfer by recognized professional authorities, for international certification of validated methods. Due to their accuracy, high sensibility, and versatility, the laser methods are recommended for remote controlled investigation, for transportable, portable, and tele-operated instrumentation. This Volume of Proceedings comprises selected contributions related laser induced phenomenas, laser investigations and documentation of most recent laboratory studies and on-site applications. For this volume I was honoured to receive a priceless and generous support from all co-editors: John F. Asmus, Marta Castillejo, Paraskevi Pouli, Austin Nevin. I Iike to put in light a very discrete and special person who supports with enthusiasm our activities from the first moment of the first project. The team of CERTO Department is thankful to Dr. Roxana Savastru for her full involvment, for never lost energy, and not less for the wonderful lesson of professional and management commitments. Roxana Radvan LACONA VIII Chair Bucharest, September 2010

ix

LACONA8.indb ix

11/10/2010 5:36:02 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Permanent scientific committee

Margaret Abraham

Los Angeles County Museum of Art, USA

John Asmus

IPAPS, University of California, San Diego, USA

Gerd v. Bally

Laboratory of Biophysics, University of Münster, Germany

Giorgio Bonsanti

University of Florence and Centro Europeo di Ricerche sul Restauro (CERR) di Siena, Italy

Marta Castillejo

Instituto de Química Física Rocasolano, CSIC, Madrid, Spain

Martin Cooper

The Conservation Centre, National Museums Liverpool, UK

Klaus Dickmann

Laserzentrum FH Münster, Germany

Costas Fotakis

Foundation for Research and Technology Hellas, IESL, Heraklion, Crete, Greece

Wolfgang Kautek

University of Vienna, Department of Physical Chemistry, Vienna, Austria

Eberhard König

Freie Universität Berlin, Germany

Mauro Matteini

Istituto per la Conservazione e Valorizzazione dei Beni Culturali, CNR, Florence, Italy

Johann Nimmrichter

Bundesdenkmalamt, Austrian Federal Office for the Care of Monuments, Centre of Art Conservation, Vienna, Austria

Roxana Radvan

National Institute of Research and Development for Optoelectronics, Romania

Renzo Salimbeni

Istituto di Fisica Applicata Nello Carrara CNR, Florence, Italy

David Saundersn

The British Museum, Department of Conservation, Documentation and Science, London, UK

Manfred Schreiner

Academy of Fine Arts Vienna, Austria

Matija Strilic

University College London—Centre for Sustainable Heritage The Bartlett School of Graduate Studies, London, UK

Veronique Verges-Belmin

Laboratoire de Recherche des Monuments Historiques, Champs-sur-Marne, France

Kenneth Watkins

Department of Engineering, University of Liverpool, UK

Vassilis Zafiropulos

Technological Educational Institute of Crete & Center for Technological Research—Crete, Sitia, Crete, Greece

xi

LACONA8.indb xi

11/10/2010 5:36:03 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Gauguin, Mucha, and Art Nouveau J.F. Asmus University of California, San Diego, La Jolla, CA, USA

ABSTRACT: The Art Nouveau movement traces its earliest stirrings to the middle of the 19th Century and several of that era’s well-known impressionist painters. In the year 1895 Alfons Mucha’s Sarah Bernhardt poster burst upon the Paris theater scene and this Art Nouveau sensation captured the public imagination. A few months earlier in Brittany, Paul Gauguin had been composing female images for Tiffany until he became bedridden in the aftermath of a fistfight with three sailors. During the twomonth convalescence Gauguin’s model, “Annah la Javanaise”, stole his belongings in order to raise money through sales in Paris. We have performed digital computer image analyses of surviving items by Mucha and Gauguin revealing that one of Gauguin’s artworks fell into Mucha’s hands and was photographically copied. Thus, Mucha’s famous Gismonda poster for the Bernhardt performance in the Sardou play had been plagiarized and Gauguin, instead, was the actual source of the Art Nouveau phenomenon. 1

INTRODUCTION

of 1894 (Reade, 1963). The manager of the Theatre de la Renaissance called Mucha’s printing shop asking if anyone there could design a poster for Gismonda as his artist was ill. It was stipulated that the poster had to be composed, executed, and published within five days as Madame Bernhardt insisted that it appear on December 31. In spite of this crushing deadline and tremendous pressure the posters were drying by December 31 and distributed throughout Paris on January 1. It was an immediate success and Parisians applauded the launching of the school of the “new art”. Soon “Style Mucha” became an international movement known as “Art Nouveau” and the unknown Czech artist, Alfons Mucha, became his native country’s national hero. The accidental discovery of the postage-stamp collage raises the specter (from historical, stylistic, and technical perspectives) of an act of plagiarism. These aspects of the issue are explored and analyzed in the forthcoming sections of this paper.

A small collage of paper fragments pasted onto the glazed face of a ceramic tile appeared in 1954 at an auction of objects from a San Marino, California estate. The collage design portrays the actress, Sarah Bernhardt, in the title role of Gismonda in Sardo’s 1894 Paris theater production. A careful inspection under magnification reveals that the figure was created by gluing snippets from at least sixty French postage stamps onto the of a common household ceramic tile. It appears that the stamps have all been cancelled, however the cancellation marks are undecipherable due to the manner in which the very tiny fragments are dispersed throughout the entire composition. No paint or ink is used to outline the lady. The borders consist of dark lines cut from postage stamps. In 1967 the tile with the attached collage was shown to the French art dealer Oscar Meyer (Tyler, 1988). He immediately exclaimed: “My God, that’s by Paul Gauguin!” He identified the Gauguin selfportrait as a jester appearing behind the banner on the plinth and the “GP” initials at the lower right-hand corner. (He had seen receipts signed by Gauguin in exactly that manner). In addition another “GP” is formed by Bernhardt’s right hand and sleeve, and various “PG” initials appear in the drapery of the gown. These discoveries planted the first seeds of what was to grow into a theory that Gauguin, not Alfons Mucha, was the authentic father of the Art Nouveau movement. According to Mucha’s son, the events leading to the overnight public infatuation with the school of Art Nouveau began during the Christmas holidays

2

LATE 19th CENTURY PARISIAN ART

French art and culture rebounded in a wave of vigor and optimism following the trauma and agony of the Franco-Prussian War, the capitulation of the Imperial Army in 1870, and the ensuing political and civil upheavals. French Impressionism became the worldwide focus of a new renaissance in arts and architecture. Artists, dealers, collectors, and bohemians flocked to Paris in order to celebrate the sense of a new and explosive vitality. The Eiffel Tower symbolized a revolutionary

1

LACONA8.indb 1

11/10/2010 5:36:03 PM

stones at Concarneau, when I was walking with Annah. I knocked down with two punches a pilot who attacked me. I took them all on, and kept the upper hand, until my foot caught in a hole and in falling I broke my leg.” He was confined to bed for two months and his inability to paint led to long hours, days, and weeks of boredom. To another friend he replied: “Your letter surprised me in utter idleness: in front of me a heap of unanswered letters growing higher each day.” We speculate that this “heap” constitutes the material of the stamp collage. Although Gauguin is best known for his easel paintings, he worked in various media including wood and metal. At the time of his stay in Brittany he was also collaborating in designs for stained glass that contributed to the art of Tiffany. Thus, creating a collage would be in keeping with his broad interests in diverse media (Hoog, 1987). During Gauguin’s convalescence it is known that he was largely limited to reading his favorite monthly publication, “Le Figaro Illustre”, which featured a color picture of Sarah Bernhardt as she appeared between acts of the play Izeyl. She was greatly admired by Gauguin, as, in addition to her acting, she was an accomplished painter and sculptress. It is doubtful that he saw her except in photographs as in the le Figaro illustration. (However, there are historical hints that Gauguin may have traveled to London a few months earlier, specifically, to witness a Bernhardt stage performance there). It is plausible that Gauguin was inspired by the Bernhardt photograph, his anticipation of her appearance in the Sardou (his idol) Gismonda production, and the widely admired pose of Liberty Enlightening (Statue of Liberty, also depicted in Le Figaro) by Bartholdi. The Gismonda collage presents itself as a synthesis of these three elements. Gauguin’s confinement, inability to paint, mountain of postage stamps, infatuation with the subject, and access to a ceramic tile (the renovation of his rooming house: Figure 3) make him a likely candidate composer of the collage. Nevertheless, this raises the speculative question as to how this artifact (if created by Gauguin) might have fallen into the hands of Alfons Mucha in Paris. Gauguin’s letters to friends suggest the following scenario. With each passing day of Gauguin’s confinement and recuperation his model and mistress, Annah la Javanese, became progressively more restless and less agreeable. She was probably bored with Gauguin and his country life. She longed to return to the excitement of Paris and did so in September with any effects of value that she was able to loot from Gauguin’s possessions. (Perhaps, the original artwork composed upon the tile was among these items). Upon Annah’s return to Paris she began modeling for Mucha, became his mistress, and took up residence in his studio (Figure 4).

Figure 1. Typical Parisian posters from the era preceding the appearance of Mucha’s Gismonda (1894) that heralded the Art Nouveau movement. On the left is a liquor advertisement (1893) by Henri Guydo. Henri de Toulouse-Lautrec produced the poster on the right to promote a performance (1892) by the cabaret singer, Aristide Bruant.

Figure 2. Sarah Bernhardt (left), whose performance in Sardo’s Gismonda inspired the emergence of Art Nouveau. One of Mucha’s most famous designs (La Plume) illustrating the pinnacle of his “Style Mucha” (right).

fusion of art and industry. Names such as Degas, Manet, Monet, Vuillard, Van Gogh, Gauguin, and Renoir became the most recognizable icons of Western Art. Henri de Toulouse-Lautrec became famous as a prolific practitioner of the “Poster Art” branch of Impression. Figure 1 reproduces images of the styles typical of the poster art before the appearance of the “New Art” (Style Mucha/Art Nouveau) of Alfons Mucha and his portrayals of Sarah Bernhardt and her theater appearances (Figure 2). 3

HISTORICAL MILIEU

In May 1894 Paul Gauguin moved from Paris to Pont-Aven in Brittany. While on a stroll with friends and fellow artists, his bizarre attire provoked a shouting match with three passing sailors. In a letter to a friend Gauguin related the climax of the encounter: “They started throwing

2

LACONA8.indb 2

11/10/2010 5:36:03 PM

4

GEOMETRICAL COMPARISON OF THE TILE AND POSTER

The famous 1894 Gismonda poster by Mucha appears to scale juxtaposed with the postage-stamp collage in Figure 5 (black and white reproductions in both instances). In determining the probable origin of the Gismonda poster design it is relevant to compare the precise geometries and details of the collage with those of the poster. In principle the flicker technique, photogrammetry, and 3-D video imaging are both convenient and informative in such comparative studies. However, for hardcopy transmission to scholars (viz., art historians) and convenient interpretation we found that a colorcoded overlay provides the most revealing display for our purposes. This may be accomplished by means of a photographic double exposure in which one image is in one color and the other is in another. Thus, when the two images match, the colors combine (e.g., red and green may combine to produce yellow). In regions where image contours are not superimposed the degree of color displacement reveals geometrical image differences. With the advent of digital computer image processing this type of operation is much more readily and precisely performed with a computer than by means of photographic reproduction. Our first step in implementing such a bicolor Gismonda superposition was the digitization of highquality photographic reproductions of the poster and collage images. A high-pass image-processing (edge detection) algorithm was applied to both images in order to enhance outlines and details for comparison. The (5 × 5) kernel in the transformation was:

Figure 3. A Pont-Aven street (left) showing the irregular pavement where Paul Gauguin may have fallen and broken his leg during a brawl with three seamen. The rooming house (right) in Pont-Aven where Paul Gauguin lived with Annah la Javanese and recuperated from his broken leg in 1894.

Figure 4. Photographic self-portrait of Alfons Mucha in his Paris apartment/studio in early 1894 (left). Mucha photograph of some friends in his studio in 1894 (right). His future model and lover, Annah la Javanese, is in the rear at the center. Gauguin is in front.

−1 −1 −1 −1 −1

The Bernhardt poster emergency fell upon Mucha’s print shop several weeks after Annah joined him in Paris. It is plausible that in the urgency of those final hours of December Mucha came upon a desperate solution to his dilemma. He would have placed the Gismonda tile in photographic enlarger, traced the poster-sized image on paper, and use the copy to prepare the lithographic press. There would have been no time or sensible rationale for seeking Gauguin’s permission for use of his pilfered property. This hypothetical scenario establishes a circumstantial case for the provenance of the Gismonda tile and its role in the genesis of Art Nouveau. However, the superficial plausibility of this explanation does not constitute a defensible proof. Toward this end, digital computer Image Processing (IP) has been applied to the study of the Gismonda image of the tile collage and the Mucha poster. We also employ IP to analyze Gauguin’s signature and compare it to scratches on a cane from the same estate lot of the tile.

−1 −1 −1 −1 −1

−1 −1 25 −1 −1

−1 −1 −1 −1 −1

−1 −1 −1 −1 −1

Figure 5. A black and white reproduction of the sensational 1894 Gismonda poster by Alfons Mucha (center). The postage stamp collage is shown (black and white) to scale on the right.

3

LACONA8.indb 3

11/10/2010 5:36:03 PM

which signifies that the intensity value for the pixel in question is multiplied by 25 and the 24 surrounding values are multiplied by −1. The central pixel value is replaced by the normalized sum of all 25 numbers. As the original two photographs from which Figure 5 was produced were of different magnifications, it was necessary to scale the digital image files to identical sizes. This was accomplished by adjusting the respective scales so that face-neck lengths were equal numbers of pixels. The horizontal scale was checked by means of the separation of the hands. The initial geometrical comparison was accomplished next by digitally subtracting the enhanced Mucha design from the enhanced collage design. The result is shown in Figure 6 (left) with the original red/green pseudo color-coding replaced by a gray scale for this publication. In Figure 6 the Mucha poster design edges are displayed in gray. The collage design edges appear as white lines. A careful inspection of the Figure 6 (left) overlay reveals that there is an almost perfect match between the outlines of the two compositions for the top halves. Specifically, by following the profiles of each palm leaf or gown detail it is seen that they are represented by light-dark line pairs that follow identical trajectories in most instances. On the other hand, when the entire superposition is considered (rather than just the top half), an entirely different picture emerges. The entire lower half exhibits a major vertical displacement. This suggests that a central horizontal strip of the Mucha poster is missing (with respect to the collage composition). In order to illustrate this hypothesis that the Mucha design is incomplete, this image was sliced at the center and the upper and lower portions were separated (in the computer’s digital

Figure 7. Typical design detail from the front of the Gismonda gown (left). The same gown design detail at the location of the apparent break at the middle of the Gismonda poster (right) identifying the elements that have been removed.

image file). Figure 6 (right) displays the consequences of such an operation. The top and bottom halves of the Mucha poster image conform perfectly to respective portions of the collage image. This shows that the poster is missing a band of detail at the center constituting about 5% of its length. This conclusively establishes that the collage cannot be a copy of the poster. On the other hand this opens the door to the possibility that the poster could be an enlarged copy of the collage. Thus, it is technically possible that the collage may be the template from which the poster design was derived. In producing the Figure 6 (right) overlay the poster image was arbitrarily sliced at the center as a cursory inspection indicated this level as the origin of the discontinuity. However, we were able to determine the precise position of the dislocation by considering design details of the gown. The front panel of the Gismonda gown has a repeating sea creature/foliage design with four cycles. Figure 7 (left) reproduces one floral element in one cycle of the complete branch with leaves. This appears in a complete form in all four cycles on the collage and three of the four cycles on the Mucha poster. However, on the poster the fourth of the cycles is presented in an abridged and unsymmetrical form (Figure 7, right). This feature appears at the second cycle from the top of the gown panel. This is at the level of the proper right hand of the female figure. This element is complete on the collage. 5

WALKING STICK PROVENANCE

We developed another piece of evidence to consider in addition to the expert attribution of the collage tile to the hand of Paul Gauguin (Section 1). There was an antique hand-carved walking stick in the estate sale with the tile collage. Clearly, Gauguin would have required such assistance once his leg had healed enough for him to leave his bed. The handle end of this hand-carved cane exhibits faint traces of deliberate scratches. A black and

Figure 6. Overlays of the edge-enhanced outlines of the Gismonda figures from the Mucha poster (gray) and the postage-stamp collage (white). The overlay on the left reveals that the poster figure is shorter in proportion than the collage. On the right the poster image has been split and the lower half has been moved downward in order to match the collage at both the top and the bottom.

4

LACONA8.indb 4

11/10/2010 5:36:07 PM

are virtually identical except for a horizontal band that is omitted at the center of the poster. As the two Gismonda artworks are so different in size, the only 19th Century technology capable of producing a scaled copy of such high fidelity would be an optical photographic enlarger such as those available to Mucha in his employer’s print shop. If one supposes that the tile collage was a photographically guided copy of the Mucha poster, there is no sensible explanation for adding a strip in the middle. Conversely, the lithographic process is such as to quite easily lose a strip in going from an enlarged image of the tile to the poster. (It is customary to print large posters with two lithographic plates and detail at the joint between the two plates is lost). A third possibility is that both pieces are copies of some common ancestor. However, on the one hand there is no evidence for such a work. On the other this would involve two independent optical copy procedures with the attendant doubling of the degradation in spatial fidelity. This is belied by the observed precision of the spatial match. Further, the mechanics of collage making are such that it would be vastly more difficult to produce from an optically projected image than would be a drawing. Lastly, evidence that the collage is by Gauguin is compelling as the autographs on both the collage and the associated walking stick conform to those known to be of the artist. The inescapable conclusion to be drawn from this study must be that Gauguin is the originator of the design of the Gismonda poster that triggered the rise of the Art Nouveau movement. What remains to be resolved is whether Gauguin (in distant Tahiti) knew of Mucha’s use of the design, whether he gave his friend permission to use it, or whether it was simply an act of plagiarism.

Figure 8. A photograph of the handle of a wooden walking stick (top) from the same estate sale as the postage-stamp collage. Computer-enhanced (FFT, edge enhancement, gain-bias adjustment) scratches on the handle of the walking stick (center). Typical Paul Gauguin signature from a painting (bottom).

white copy of a color photograph of this cane handle is reproduced at the top of Figure 8. We employed computer image enhancement to each of the scratches in an effort to decipher them. We employed FFT and Hi-Pass spatial filters as well as pseudo-color and gain-bias transformations. The results of these numerical operations are shown at the center of Figure 8. Gauguin’s writing style was notably idiosyncratic so it seemed implausible to suppose that our performing cross correlations with diverse alphabet styles would prove fruitful. Consequently, we relied on a subjective visual comparison of the enhanced scratches with a characteristic Gauguin signature (Figure 8, bottom). We concluded that the walking stick scratches are probable representations of Gauguin’s signature. It is known that Gauguin worked with several different media in his art. It follows that it is plausible to surmise that he may have carved the walking stick himself. In late 1894 when Annah left Gauguin and stole his property she probably realized that his art was beginning to attain some marketable value (as did one of his landladies who refused to relinquish some of his paintings). It seems that Annah pilfered both the cane and the Gismonda collage from Gauguin’s possessions and took them to Mucha’s studio when she moved in with him. Then Mucha had a perfect opportunity to copy Gauguin’s collage composition under the emergency situation of having only days to create a commercial poster for Bernhardt. The association of the collage with the walking stick supports this scenario. 6

ACKNOWLEDGEMENT Mr. Charles Tyler of Brentwood, California proposed and developed the Art Nouveau genesis theory summarized above. He also paid the mainframe computer (CRAY) time charges associated with the digital image manipulations of this investigation.

REFERENCES Hoog, M. 1987. Paul Gauguin Life and Work. New York: Rizzoli. Reade, B. 1963. Art Nouveau and Alphonse Mucha. London: Her Majesty’s Stationery Office. Tyler, C. 1988. An Art Nouveau Phenomenon. Los Angeles: Charles B. Tyler Publisher.

CONCLUSIONS

In conclusion, it emerges from this computer-aided comparison of the two designs that the geometries

5

LACONA8.indb 5

11/10/2010 5:36:07 PM

Innovative approaches in laser cleaning researches and instrumentation development

LACONA8.indb 7

11/10/2010 5:36:07 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

The effect of ultrafast lasers on laser cleaning: Mechanism and practice K.G. Watkins & P.W. Fitzsimons Laser Group, Department of Engineering, University of Liverpool, Liverpool, England, UK

M. Sokhan Department of Conservation, City & Guilds Art School, London, UK

D. McPhail Department of Materials, Imperial College, London, UK

ABSTRACT: The increased availability of picosecond (ps) and femtosecond (fs) laser sources opens the prospects of new tools for art restoration. However, the use of such ultrafast lasers has most often been investigated in application areas such as material removal by ablation which is more proper to engineering requirements rather than conservation. The purpose of this investigation is to consider the effectiveness of these ultra fast laser types under the specific requirements of the conservation activity and to determine whether new mechanisms leading to strictly limited material removal on the nano and micro size scales are operational. A dielectric layer of TiO2 is considered and the results are analysed by means of surface analysis techniques in order to deduce the operative mechanism in each case and to assess the applicability of ultra short pulse lasers in art conservation. 1

could play an increasing role in art conservation (S. Barcikowski et al., 2006; A.V. Rode et al., 2006; S. Georgiou et al., 2008; S. Gaspard et al., 2008). As conservators use these laser types, what mechanisms will be involved? Will the mechanisms be the same, given that the pulse length is now much shorter (200 fs–20 ps compared with

INTRODUCTION

Most conservation work with lasers is approached empirically. However, there are good reasons to enquire about what mechanisms are operating when material is removed from the artefact surface. Surprisingly, there is still a shortage of knowledge about what is actually taking place when an art object is cleaned by laser. For the case of the laser types with pulse length varying from continuous wave down to nanosecond pulses that are most often currently employed, it is clear that more than one mechanism is involved (K.G. Watkins et al., 2005). As shown schematically in Figure 1, a range of mechanisms can be employed depending on the laser absorbed intensity and the pulse length. Lasers with much shorter pulse length— picosecond and femtosecond lasers—have been available in research laboratories and in industry for some time but these are bulky, expensive and mainly unavailable in art conservation. Very recently, new, low cost, small size picosecond and femtosecond lasers have become available that could extend the tool set available for conservation. The short pulse passively mode locked fibre lasers that are becoming available are of small form and of low cost compared with earlier short pulse laser types and

Figure 1. Absorbed intensity versus interaction time diagram showing schematic regimes of candidate laser cleaning mechanisms (K.G. Watkins et al., 2005).

9

LACONA8.indb 9

11/10/2010 5:36:07 PM

to stop at the layer—substrate interface, it could be self limiting. A key advantage is that the extent of material removal could be controlled down to the nanometer scale. This would amount to a new process of ultra-fine laser cleaning. Because the pulse repetition rate is so large in these new ultrashort pulse laser types, cleaning rate in terms of area coverage (as the laser is rastered over the work piece) is expected to be high. Stoian et al. (2008), investigating laser ablation of Al2O3 by a coulombic repulsion mechanism with 100 fs laser pulses, found an initial gentle ablation regime in which a ‘few nanometers’ is removed per laser shot, followed by a strong ablation regime characterised by an order of magnitude increase in ablation rate. The crossover point was after 20 laser pulses, the transition attributed to a buildup of mechanical deformation (F. Aumayr et al., 1999). Both regimes may be valuable in art conservation. The objective of this preliminary study is to establish evidence for the controlled removal on a nanometer scale of material from titania by the use of femtosecond laser pulses and to offer this as a contribution to the recognition of the new laser types as valuable new tools in art conservation. The text should fit exactly into the type area of 187 × 272 mm (7.36" × 10.71"). For correct settings of margins in the Page Setup dialog box (File menu) see Table 1.

10 ns for Nd:YAG) and the repetition rate much higher (up to 100 MHz compared with 10 Hz for Nd:YAG)? At these very short pulse lengths, laser intensity at the treated surface will be very large −1010–1014 W/cm2. This compares with laser intensity at the treated surface of 107–109 W/cm2 for typical Nd:YAG cleaning. At the high intensity produced by the ultrashort pulse laser types, new mechanisms become possible. This may lead to new forms of controlled removal of material from a surface. The mechanisms involved are likely to be as complex in scope and scale as those summarised for pulse lengths down to nanoseconds, as shown in Figure 1, and it is unwise at this stage to be restrictive. It is more proper to consider candidate mechanisms. One such candidate is the coulombic repulsion mechanism (R. Stoian et al., 2002). The key concept is that under highly intense laser radiation, the highly energetic photons can remove electrons from the bound atoms or molecules in the surface of the materials and the atoms / molecules there will obtain a net positive charge. If the resulting repulsive force is large enough to overcome the bonding forces, material will be removed by coulombic repulsion. Note that there would be no melting; the process of removal would be more akin to sublimation (phase change without melting). In most cases in art conservation, avoidance of melting is desirable. More than one photon may be required to promote an electron from the valence band to the conduction band (multiphoton ionisation). To achieve this laser intensity approaching 1012 W/cm2 is required. Coulombic repulsion is unlikely to be active in metals (J. Cheng et al., 2009). Metals have abundant free electrons that would simply replace the electrons removed by ionization. However, in art conservation, interest is normally in removing non metals—even on metals it is the oxides and sulphides that require removal. If the process was

Figure 2.

2

EXPERIMENTAL

The laser used in this study was a Clarke MXR CPA 2010 1 W average power femtosecond laser with a 1 kHz repetition rate operating at a wavelength of 387 nm. The output pulse energy was controlled via an external attenuator and ranged between 0.638–3.46 μJ with a temporal pulse

Schematic diagram of the experimental arrangement.

10

LACONA8.indb 10

11/10/2010 5:36:08 PM

Table 1.

Depth of ablation (nm) of TiO2 with fs pulses as a result of varying the fluence and pulse number. Number of pulses

Fluence (J/cm2)

0

5

10

15

20

30

40

50

0.32 0.48 0.72 1.02 1.29 1.73 2.15

0 0 0 0 0 0 0

– – 21.20 26.24 38.62 45.48 38.30

– – 24.40 32.66 41.80 45.80 42.24

– 29.74 25.18 36.98 41.70 46.74 –

15.90 31.4 27.80 40.02 45.72 – –

31.60 38.08 28.76 44.98 43.24 – –

37.80 34.36 30.82 46.68 X – –

34.50 39.96 41.44 X X – –

Legend: no results.

This shows that it is possible using fs pulses to ablate material from the TiO2 layer on a nanometre scale. The minimum depth recorded was 15.90 nm (recorded at 0.32 J/cm2 and 20 pulses) and the maximum was 46.74 nm (recorded at 1.73 J/cm2 and 15 pulses). Significantly all the material removal was less than 100 nm in depth, showing the possibility of ultrafine laser cleaning. At fluences of 0.32 and 0.48 J/cm2 the onset of surface ablation did not occur until 15 and 10 pulses, respectively; this indicates that the number of pulses is a contributing factor in the ablation process. As the surface was exposed to successive pulses the ablation threshold of the TiO2 was altered; so that after an incubation period ablation was initiated at the surface. Figure 3 is a white light interferometric measurement of the TiO2 surface post treatment; all depth measurements were made on the Y profile (Figure 3c). The average depth was recorded with respect to the original layer height. This measurement was taken at the lowest fluence recorded and shows that 50 pulses achieved an average ablation depth of 34 nm; this corresponds to an average ablation depth per of 0.68 nm per pulse if the incubation period was not considered. However, it was observed that the surface roughness at the centre of the ablated region is very high; this type of surface quality post processing would not be ideal for conservation needs. Figure 4 shows the ablated region after the surface had been exposed to 15 pulses at 1.73 J/cm2; the quality of the ablated surface was significantly improved over the surface quality observed at 0.32 J/cm2. The average depth of ablation recorded was also 34 nm; however this was achieved at significantly lower pulse numbers than Figure 3. The average depth of ablation recorded per pulse was 2.2 nm. It is immediately apparent that the surface quality in the ablated area is much improved over those observed at the lower fluence (0.32 J/cm2); surfaces of this quality would be much more applicable to conservation.

length of 180 fs (FWHM). The number of pulses was controlled using a shutter with an accuracy of ±3 ms. The focussed spot size was determined to be 29 μm. A 3-axis motion control system (Aerotech) was used in conjunction with NView MMI software to manipulate samples. TiO2 samples were prepared by sol-gel deposition onto a glass substrate via spin coating. A sol-gel of TiO2 was synthesised and added dropwise onto a silicon wafer. The sample was spun at 2000 rpm for one minute. An amorphous film was formed by heating the sample at 80°C for ten minutes and then at 320°C for a further thirty minutes. The process of adding the solution drop wise and heating was repeated until the desired number of layers had been added; at this point a final heat treatment of 500°C for sixty minutes was carried out to produce a consolidated film. The depth of the applied layer was measured to be approximately 500 nm. A single sample of TiO2 was exposed to an increasing number of pulses (5, 10, 15, 20, 30, 40 and 50) at increasing fluences (0.32, 0.48, 0.72, 1.02, 1.29 and 1.73 J/cm2). Through variation of these parameters and recording the average depth of ablation observed, it was possible to tabulate and quantify the effect of both the fluence and number of pulses upon the TiO2 layer. Surface analysis and depth profiling was performed using an optical microscope (Nikon) attached to a CCD camera and white light interferometry (WYKO NT1100). A Phenom (FEI) Scanning Electron Microscope (SEM) system was utilised to provide a highly detailed image of the ablated region. 3

RESULTS AND DISCUSSION

Table 1 shows the depths of ablation recorded on the single TiO2 sample with increasing fluence and pulse number; all observed ablation depths are given in nanometres.

11

LACONA8.indb 11

11/10/2010 5:36:08 PM

Figure 3. Ablated region of TiO2 surface. A fluence of 0.32 J/cm2 and 50 pulses were used. From top left clockwise: a) Topographic image; b) cross sectional profile in the X-plane; c) cross sectional profile in the Y-plane including depth measurement; d) 3D image of ablated region.

Figure 4. Ablated region from TiO2 surface. A fluence of 1.73 J/cm2 and 15 pulses were used. From top left clockwise: a) Topographic image; b) cross sectional profile in the X-plane; c) cross sectional profile in the Y-plane including depth measurement and d) 3D image of ablated region.

12

LACONA8.indb 12

11/10/2010 5:36:08 PM

calculated to be between 5.36 × 1011 and 1.81 × 1012 W/cm2. At intensities of around a terawatt, the optical properties of materials can be altered, leading to nonlinear absorption (R.L. Sutherland et al., 2003). This effect is characterised by a well defined ablated region with little or no damage to the immediate surrounding area. From the cross sectional profiles provided by white light interferometry and the SEM images, it was observed that the ablated region matches this expected criteria and that nonlinear absorption is likely. The mechanism for material removal utilising ultra short pulse lengths would not proceed under the same conditions observed at longer pulse durations (i.e. nanosecond). In work on alumina reported by Ashkenasi et al. (2000), which utilises intensities and pulse lengths similar to those presented here, the ablation process observed was coulombic repulsion, as determined by the detection of fast ions using mass spectrometry and the transition from a weak to a strong ablation regime (R. Stoian et al., 2000; R. Stoian et al., 2001; D. Ashkenasi et al., 2000) where the depth of ablation increases significantly. The mechanism requires the high optical intensities (A.V. Rode et al., 2006) provided by ultrashort pulses. To confirm the presence of this mechanism in the present case requires further study, specifically of the ablated matter produced in the first few moments after ablation.

Figure 5 shows the variation of ablation depth with number of pulses. It can be seen that generally as the number of pulses and fluence increases so did the depth of ablation observed. However, the depth of ablation observed for increasing number of pulses at 0.72 J/cm2 did not follow the expected trend; the observed values at this fluence were lower than the majority of the results recorded for 0.48 and 0.32 Jcm2. This could be due to the large variation in surface roughness observed at 0.32 J/cm2. In the white light interferometry results, measurement of the depth of ablation was not taken from the lowest point recorded by the interferometer. Two artefacts at the edge of the ablated region which rise above and below the surface are visible (see Figure 6); these features are not present on the surface. This was shown to be the case by viewing the ablated region using an SEM in backscattered electron imaging mode (Figure 7). The debris on the surface of the sample is due to the fracture process required to examine the surface under SEM. It is not possible to describe an exact mechanism of ablation based on the recorded results. However, it is possible to suggest a mechanism based on the pulse length, intensity and images of the surface. From the recorded pulses energies and the measured spot size, the intensity on the surface is 60

50

40

30

20

10 0

10

20

30

40

50

60

Figure 5. Variation of ablation depth with increasing pulse numbers. Legend shows the Fluence (J/cm2).

Figure 7. SEM images of the ablated region. The top row shows the ablated holes under full illumination at a low (left) and high (right) magnification. The bottom row consists of a high magnification hole under topographic illumination.

Figure 6. Expanded X-axis profile from Figure 4 highlighting the artefacts.

13

LACONA8.indb 13

11/10/2010 5:36:09 PM

4

CONCLUSIONS

F. Aumayr, J. Burgdorfer, G. Hayderer, P. Varga and HP. Winter, “Evidence against the “Coulomb Explosion” Model for Desorption of Insulator Surfaces by Slow Highly Charged Ions”, Physica Scripta, Vol. T80, 240–242, 1999. J. Cheng, W. Perrie, B. Wu, S.P. Edwardson, M. Sharp, G. Dearden, and K.G. Watkins, “Ablation Study on Metallic Materials with a Picosecond Laser: Experimental and Simulation Analysis”, 2009. K.G. Watkins, and W.M. Steen, “Laser Materials Processing, Chapter 9: Laser Cleaning”, Third Edition, Springer-Verlag, 2005. R. Stoian, A. Rosenfeld, D. Ashkenasi and I.V. Hertel, “Surface Charging and Impulsive Ion Ejection during Ultrashort Pulsed Laser Ablation”, Phys Rev Letters, Vol. 88, 9, 2002. R. Stoian, D. Ashkenasi, A. Rosenfeld and E.E.B. Campbell, “Coulomb Explosion in Ultrashort Pulsed Laser Ablation of Al2O3”, Volume 62, Number 19, Physical Review B, 2001. R.L. Sutherland, “Handbook of Nonlinear Optics” 2nd Edition, Marcel Dekker, 2003. S. Barcikowski, N. Barsch, T. Burmester, J. Bunte, J. Ulrich, A. Gervais and M. Meier, “Femtosecond Laser Cleaning of Metallic Antique Artworks– Advantages, Limits and Economic Aspects”, Laser Cleaning II, D.M. Kane, World Scientific Publishing Ltd. 2006. S. Georgiou, D. Anglos and C. Fotakis, “Photons in the service of our Past: Lasers in the preservation of Cultural Heritage”, Contemporary Physics, Vol. 49, No. 1, 1–27, 2008. S. Gaspard, M. Oujja, P. Moreno, C. Mendez, A. Garcia, C. Domingo, and M. Castillejo, “Interaction of Femtosecond Laser Pulses with Tempera Paints”, Appl. Surf. Sci. 255, 2675–2681, 2008. X.C. Wang, G.C. Lim, H.Y. Zheng, F.L. Ng, W. Liu, and S.J. Chua, “Femtosecond pulse laser ablation of Sapphire in ambient air”, Appl. Surf. Sci. 228, 221–226, 2004.

Using fs pulses it is possible to remove nanometre layers from the surface of a dielectric through ablation. Removal of material on such a scale holds many potential benefits for conservation; i.e. the removal of over paints, tarnish layers and oxides from metals with no damage to the underlying substrate. Ablation with fs pulses produced well defined removal with no lateral damage to the surrounding surface. It is suggested that the candidate mechanism of ablation operative with the parameters used in this study is coulombic repulsion; however more experimentation is needed to confirm this. The number of pulses impinged on the surface plays an important role in the ablation mechanism. From Table 1, it can be seen that ablation can be initiated at higher pulse numbers when previously there was no ablation; this indicates that the ablation threshold of the target is altered by successive pulses. This effect is also present at higher fluences where the number of pulses influences the total depth of ablation. REFERENCES A.V. Rode, N.R. Madsen, E.G. Gamaly, B. Luther-Davies, K.G.H. Baldwin, D. Hallam, A. Wain and J. Hughes, “Ultrafast Laser Cleaning of Museum Artefacts”, Laser Cleaning II, D.M. Kane, World Scientific Publishing Ltd. 2006. D. Ashkenasi, R. Stoian, and A. Rosenfeld, “Single and multiple ultrashort laser pulse ablation threshold of Al2O3 (corundum) at different etch phases”, Appl. Surf. Sci. 154–155, 40–46, 2000.

14

LACONA8.indb 14

11/10/2010 5:36:09 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Spectral analysis of the effects of laser wavelength and pulse duration on tempera paints M. Oujja & M. Castillejo Instituto de Química Física Rocasolano, CSIC, Madrid, Spain

P. Pouli & C. Fotakis Institute of Electronic Structure and Lasers (IESL), Foundation for Research and Technology-Hellas (FORTH), Crete, Greece

C. Domingo Instituto de Estructura de la Materia, CSIC, Madrid, Spain

ABSTRACT: The application of laser cleaning methodologies on light-sensitive cultural heritage substrates (paintings, documents on paper and parchment, textiles, etc) requires the study of the physicochemical effects that may be induced upon laser irradiation to the constitutive materials. We present here work carried out to analyze the influence of laser wavelength and pulse duration on the modifications induced on egg yolk based tempera paints by using pulses of 150 picoseconds (at 1064 and 213 nm) and 10 nanoseconds (at 213 nm) on unvarnished aged model samples of unpigmented and artist’s pigment temperas (vermillion and lead chromate). A multianalytical approach was chosen based on the use of colorimetry for the quantification of colour changes, and spectrofluorimetry and Fourier Transform Raman spectroscopy (at 1064 nm) to assess possible laser induced chemical changes. An important observation from these experiments is that discoloration at 213 nm/150 ps is limited, fact that gets particularly important in the case of vermillion which is a well known laser-sensitive pigment. Raman bands attributed to the pigment in the tempera sample remain unchanged upon laser irradiation, except in the case of vermillion where the pigment bands tend to disappear from the spectra. Absence of amorphous carbon bands rule out carbonization or charring of the paint layer upon irradiation. Comparison of the obtained results using the pulse durations and wavelengths of this study with those previously available obtained with UV and IR nanosecond pulses illustrate the participation of mechanisms of diverse origin according with the pigment chemical nature and highlights the importance of the optimization of the laser parameters, mainly fluence and wavelength, in conservation treatments. 1

INTRODUCTION

(Athanassiou et al. 2000, Castillejo et al. 2001, 2002, 2003a & 2003b, Cooper et al. 2002, Chappé et al. 2003, Gordon Sobott et al. 2003, Pouli et al. 2000, 2001 & 2003, Teule et al. 2003, Zafiropoulos 2003). Furthermore, it was noticed that both the raw material (Pouli et al. 2001) and the paint system are affected (Weeks 1998), while the composition of the pigment and/or the irradiation parameters may affect the extent of the induced discoloration. It is interesting to note that, although verdigris was found to be laser insensitive (Castillejo et al. 2003, Pouli et al. 2001), vermillion (which in any case is one of the most known light-sensitive pigments) was found to discolour, in all studied wavelengths, at relatively low fluence values. Initially, given the sensitivity of paint systems to heat, photo-thermal phenomena were considered to explain the discoloration. This hypothesis was soon reconsidered as

This study aims to re-assess, on the basis of new advances in laser technology, the laser induced discoloration of pigments, which is considered an important drawback for the use of lasers in the conservation of paintings and polychromes. Since the early times of the laser cleaning applications (Weeks 1998) the scientific community was concerned with the sensitivity of pigments to laser irradiation and in this respect, a series of studies have been focused on the understanding of the phenomena related to the laser induced pigment discoloration. Experimentation using UV, VIS and IR laser wavelengths on a wide range of pigments both in raw form as well as in paint-systems (in various binding media) indicated that darkening of the pigment particles occurs in all the investigated wavelengths

15

LACONA8.indb 15

11/10/2010 5:36:09 PM

with a mixture of chalk and gypsum. They were prepared according to the procedure described elsewhere (Castillejo et al. 2002). Two different pigments, widely used in artistic painting practice with various chemical characteristics, were selected. These are vermillion (HgS) and lead chromate (PbCrO4), a traditional red pigment and yellow modern pigment respectively. Samples of unpigmented tempera (egg yolk) were also prepared to perform a comparison with results obtained in the binding medium itself.

reduced states of pigment compounds have been detected by means of surface specific analysis (i.e. X-Ray Photoelectron Spectroscopy, XPS) (Chappé et al. 2003, Pouli et al. 2001, Teule et al. 2003). Reduction mechanisms were further supported by the fact that darkening of several pigments (i.e. lead pigments (Pouli et al. 2001, Cooper et al. 2002)) was reversed in oxygen-rich environments and oxidised states of the laser induced products were detected. The introduction of ultra-short laser pulses of picosecond (ps) and femtosecond (fs) pulse duration in the field significantly enhanced the laser restoration possibilities, as they were found to overcome many disadvantages of the nanosecond (ns) laser pulses. Their superiority as regards the minimisation of photo-thermal, photo-mechanical and photo-chemical phenomena, independently of the optical properties of the treated material, and the optimization of morphological aspects, was shown on a variety of cultural heritage materials (Andreotti et al. 2006, Bartoli et al. 2006, Burmester et al. 2005, Gaspard et al. 2008a & 2008b, Pouli et al. 2007 & 2008) including paint systems (Castillejo et al. 2002). Initial studies on tempera paints using pulses of 120 fs at 795 nm (Gaspard et al. 2008a & 2008b) have shown the high degree of control that may be achieved in comparison to ns UV (248 nm) pulses. In continuation of this work, further experimentation on the same type of technical samples was undertaken by employing IR (1064 nm) pulses of 150 ps and UV (213 nm) pulses of 15 ns and 150 ps pulse duration. We present here results obtained in a selection of systems (Oujja et al. 2010), unpigmented, vermillion and lead chromate using colorimetry, spectrofluorimetry and Fourier Transform (FT) Raman spectroscopy, chosen to assess the physical and chemical modifications induced upon laser irradiation. Given the previous experience from the laser irradiation of the same type of samples (excimer laser at 248 nm, 25 ns pulse duration (Castillejo et al. 2002 & 2003a) and Ti:Sapphire laser at 795 nm, 120 fs pulse duration (Gaspard et al. 2008a & 2008b)), this study is expected to discuss the role of the operative wavelength and pulse duration on the laser irradiation of egg-yolk based paints, and thus to significantly approach the discoloration issue indicating the laser parameters that may overcome such effect. 2 2.1

2.2

Laser irradiation

Laser irradiation of the unvarnished tempera samples was carried out using two Nd: YAG laser systems. Pulses of 150 ps at 1064 and 213 nm were delivered by an EKSPLA, SL-312 system, while pulses of 15 ns at 213 nm were produced by a Lotis II, LS-2147 system. Irradiation tests were performed on a singlepulse basis for various fluence values (F), in order to determine ablation (Fth) and discoloration thresholds, as well as on a scanning basis in order to achieve a homogeneously irradiated area on which further analysis could be performed. Once the ablation and discoloration thresholds were determined for each tempera paint at the two irradiation wavelengths and pulse durations studied, a set of fluence values ranging below and well above the ablation threshold fluences was chosen for the scanned areas. Figure 1 shows images of the areas created on the tempera paints by irradiation at 1064 and 213 nm (with pulses of 150 ps). The irradiated surfaces (approximately 10 × 10 mm2) were created upon scanning of the beam (focused by a cylindrical planoconvex quartz lens of f = 150 mm to a spot about 10 × 1 mm2) along its largest spot dimension, with a 5% overlap between successive pulses. 2.3

Analytical techniques

Physicochemical changes induced by laser irradiation were assessed by using a combination of spectroanalytical techniques. Spectrocolorimetry served to measure the change of colour, while spectrofluorimetry and FT-Raman spectroscopy provided information on chemical modifications induced on the irradiated surfaces. A Minolta CM 2002 spectrocolorimeter served to measure the chromatic properties of the samples and specifically the changes induced by laser irradiation. Five spectra were obtained in each zone and averaged to obtain one data point. The CIELab colour space was used to measure colour shifts expressed in three variables, namely, ΔL*, Δa* and Δb*. The magnitude of the overall colour change is given by ΔE* = [(ΔL)2 + (Δa)2 + (Δb)2]1/2.

EXPERIMENTAL Samples description

The samples used for this investigation consist on unvarnished egg-yolk based tempera paint (thereafter tempera paints) applied on white panel constituted by a commercial card substrate primed

16

LACONA8.indb 16

11/10/2010 5:36:09 PM

intervals of 1 nm and excitation/emission spectral slits of 2.5 nm. All the samples were analysed at 30º from the sample axis. The emission spectra presented were taken with the excitation wavelength of 350 nm. FT-Raman spectra were recorded with a RFS 100/S–G Bruker spectrometer equipped with a cooled Ge detector. The excitation source consists in a Nd:YAG laser emitting at 1064 nm. Low laser power outputs, in the range of 10–20 mW, were used to prevent damage to the samples. Each data point was the result of the accumulation of 200 scans. The wavenumber resolution was 8 cm–1. 3 3.1

RESULTS Discoloration and ablation thresholds

The discoloration thresholds were determined by measuring the energy at which darkening of the irradiated area is observed under the optical microscope and are reported in Table 1. On the other hand, ablation thresholds were calculated by applying the spot regression method (Liu 1982). For laser pulses with a Gaussian spatial beam profile, the maximum laser fluence, F, on the sample surface and the diameter, D of the ablated area are related by D2 = 2ω0ln(F/Fth), where ω0 is the 1/e2 radius of the Gaussian beam distribution and Fth the ablation threshold. The diameter of the ablated area was determined as observed by optical microscopy. From a plot of D2 versus ln E, Fth and ω0 can be determined. Ablation thresholds of the different paint systems are reported in Table 1. It is observed that the thresholds of the pigmented systems are lower than those of the unpigmented tempera due to the fact that the effective absorption of the paint layer increases in the presence of a pigment. At a fixed wavelength (213 nm), the fluence thresholds are reduced by a factor of 1.3–1.8 when using shorter pulses of 150 ps. It is also observed that the ablation thresholds of the different systems are higher at 1064 nm than at 213 nm, due to the increase of the effective absorption of both the binding Figure 1. Tempera paint samples with zones irradiated at different fluences with 150 ps pulses: a unpigmented; b vermillion and c lead chromate (numbers near irradiated zones correspond to F/Fth). For each pigment, the upper and lower rows display areas irradiated at 1064 and 213 nm respectively.

Table 1. Ablation and discoloration (in parentheses) thresholds in mJ/cm2 for the treated tempera paints. The estimated errors are around 10%. (*) No discoloration observed. Wavelength

Spectrofluorimetric measurements were performed with a scanning Jobin-Yvon (FluoroMax-4) system with excitation provided by a Xenon arc lamp. The scanning was carried out with an integration time of 0.2 s per point in

213 nm 1064 nm

Pulse UnpigLead duration mented Vermillion chromate 15 ns 150 ps 150 ps

450 (*) 250 (*) 800 (*)

260 (*) 160 (*) 400 (70)

250 (60) 150 (50) 400 (80)

17

LACONA8.indb 17

11/10/2010 5:36:09 PM

medium and the used pigments upon irradiation in the UV region (Castillejo et al. 2002 & 2003a, Nevin 2008). 3.2

of ΔE* even at fluences below ablation thresholds. For vermillion, overall colour changes of ΔE* = 35 and 40 were measured in areas irradiated at fluences of 0.65 Fth and 1.25 Fth respectively. Under irradiation below ablation threshold, we observed darkening of the red colour and above threshold, the pigment acquires a grey metallic aspect. For the lead chromate based paint, irradiation at 1064 nm induces an appreciable degree of discoloration both at fluences below and above ablation threshold (ΔE* = 13 and 19 for fluences of 0.36 Fth and 1.1 Fth respectively). The irradiation of the unpigmented tempera paints with pulses of 150 ps at 1064 nm results in different discoloration degrees. However, examination under the optical microscope indicates that in this system the observed discoloration is due to colour changes induced to the underlying panel.

Colorimetry

Colorimetric measurements were performed on the different non-irradiated and irradiated tempera paint areas. Table 2 displays the values of ΔE* as a function of the irradiation fluence. The overall colour change of the paints irradiated at 213 nm with 150 ps and 15 ns pulses is limited in the explored fluence range, with values of ΔE* ≤ 10. Lead chromate displays the highest degree of discoloration under these irradiation conditions. In particular, the discoloration induced in lead chromate by irradiation with 213 nm, 150 ps pulses at a fluence of 1.06 Fth is ΔE* = 10. Picosecond irradiation at 1064 nm of the vermillion and lead chromate paints yields high values

3.3

Table 2. Irradiation conditions of tempera paints. F and Fth indicate the fluence used to create the irradiated areas and the ablation threshold measured for each system respectively. ΔE* indicates the overall colour change. Pigment

Wavelength and pulse duration

213 nm, 15 ns

Unpig mented

213 nm, 150 ps

1064 nm, 150 ps

213 nm, 15 ns

Vermillion

213 nm, 150 ps

1064 nm, 150 ps

213 nm, 15 ns

Lead chromate

213 nm, 150 ps

1064 nm, 150 ps

F/Fth

ΔE*

0.52 1.05 2.00 0.13 0.32 1.62 0.35 0.71 0.88 1.88

2.0 2.7 3.2 1.8 2.3 1.7 8.3 14.6 17.6 –

0.83 1.66 2.00 0.12 0.31 1.56 0.15 0.30 0.65 1.25

1.4 3.1 3.5 1.2 2.5 3.0 29.8 35.0 35.4 40.6

0.50 1.17 2.00 0.02 0.20 1.06 0.16 0.36 1.11

1.2 2.8 8.0 1.2 3.3 10.0 5.3 13.2 19.8

Spectrofluorimetry

The fluorescence spectrum of the unpigmented system consists of a broad feature extending from 375 to 650 nm (Figure 2a). This wide emission is the resulting contribution of different fluorofores that participate in the composition of egg yolk (Gaspard et al. 2008, Lakowicz 2006, Mills et al. 1994). Upon excitation at 350 nm, the fluorescence from the three aminoacids of proteins, tyrosine, tryptophan and phenylalanine, is not excited, as this wavelength is above their absorption maxima. Therefore the contribution of these compounds to the observed fluorescence band can be ruled out. In fact, in the region below 500 nm, the emission is due to the products of photo-oxidation, combination and modification of aminoacids, such as dityrosine, 3,4 dihydroxyphenylalanine (DOPA) or N-formylkynurenine (NFK) and kynurenine (Gaspard et al. 2008, Nevin et al. 2006a & 2006b, Palumbo et al. 2004, Wisniewski et al. 2007). Dityrosine, a dimeric species of tyrosine, has an emission maximum at 410 nm (Gaspard et al. 2008), while DOPA, another photo-oxidation product of tyrosine displays a characteristic emission at 480 nm (Palumbo et al. 2004, Wisniewski et al. 2007). NFK and kynurenine are products of the oxidation of tryptophan and emit at around 435 nm (Nevin et al. 2006a & 2006b). At longer wavelengths, above 520 nm, phospholipids are the main species that contribute to the fluorescence emission observed (Palumbo et al. 2004). Fluorescence spectra of the unpigmented paint recorded on areas irradiated at 213 (with pulses of 150 ps and 15 ns) and 1064 nm (with pulses of 150 ps), as shown in Figure 2a, reveal the changes induced by laser irradiation. A relative increase of intensity in the region of 400–480 nm is observed,

18

LACONA8.indb 18

11/10/2010 5:36:10 PM

pigment, which prevents the photo-oxidation of the compounds constituting the binding medium. The fluorescence spectra recorded in nonirradiated areas of lead chromate tempera (not displayed in the figure) contain two bands. The broad one, ranging from 375 to 525 nm attributed to the binding medium and the intense narrow one centred at 557 nm is assigned to the lead chromate pigment (emission due to the chromate ion chromophore, CrO42−) (Castillejo et al. 2002 & 2003a). No changes were observed for this paint system upon irradiation at 213 nm (with pulses of 150 ps and 15 ns) or 1064 nm (with pulses of 150 ps), even with fluences above ablation threshold. 3.4

FT-Raman spectroscopy

Figure 3 shows the FT-Raman spectra of the paint systems, once the spectrum of the underlying panel was adequately subtracted. Figure 3a display the spectra of the unpigmented system showing the characteristic bands of: tryptophan at 767 and 873 cm−1, the aminoacid phenylalanine at 1003 cm−1, the methylene groups of lipids and amino acids at 1302 and 1445 cm−1, the amide III and amide I of the proteins backbone at 1240 and 1656 cm−1, the C = O stretching at 1744 cm−1 assigned to fatty acid esters and the aliphatic compounds in the 2750–3100 cm−1 region (assigned to the ν(C-H) mode) (Bell et al. 1997, Drake et al. 2004, Nevin et al. 2007, Osticiolo et al. 2008, Vandenabeele et al. 2000). Irradiation at 213 nm (either with pulses of 150 ps or 15 ns) of the unpigmented tempera induces a slight decrease in the intensity of bands in the 600–1750 cm−1 region (Figure 3a), in particular those assigned to tryptophan, phenylalanine, amide III and I and fatty acid esters. However, a mild increase in the intensity of aliphatic compounds (2750–3100 cm−1 region) is observed. Irradiation at 1064 nm (150 ps) induces similar effects to those detected after irradiation at 213 nm. The decrease in the intensity of the bands in the region 600–1750 cm−1 is indicative of the degradation of aminoacids. The modifications induced by laser irradiation of the vermillion paint at 213 nm (with pulse duration of 150 ps and 15 ns), as reported in Figure 3b, show an insignificant decrease in the intensity of the pigment bands observed at 252, 282 and 343 cm−1 (Burgio et al. 2001). However irradiation at 1064 nm (150 ps) induces the total disappearance of the pigment bands indicating the degradation of the pigment. FT-Raman spectra of the lead chromate tempera do not show significant modifications upon laser irradiation at 213 nm (150 ps and 15 ns) and 1064 nm (150 ps) (Castillejo et al. 2002) in

Figure 2. Normalized fluorescence spectra taken at non-irradiated and laser irradiated areas of tempera paints: a unpigmented and b vermillion. The spectra were taken with 2.5 nm resolution at the excitation wavelength of 350 nm.

indicating the corresponding increase of photodegradation products emitting in this region. Figure 2b shows the fluorescence spectra recorded in areas of the vermillion system which are mainly pictured by two broad bands. The first one, in the region from 375 to 525 nm, is attributed to the binding medium while the second one, centred at 607 nm, is assigned to luminescence of the HgS semiconductor (Castillejo et al. 2002 & 2003a). Irradiation in the UV at 213 nm with pulses of 150 ps or 15 ns does not modify the emission spectra. However irradiation at 1064 nm in the ps regime results in the decrease of the emission corresponding to the pigment at fluences below ablation threshold and eventually its disappearance above ablation threshold, indicating that the pigment is chemically degraded. Contrary to what is observed in the unpigmented system at this wavelength, the fluorescence band of the binding medium in the painting mixture remains unaltered, due to the high absorption of laser light by the vermillion

19

LACONA8.indb 19

11/10/2010 5:36:10 PM

(Gordon Sobbot et al. 2003, Teule et al. 2003). In particular, in previous works, samples similar to the ones treated here were irradiated with KrF excimer laser pulses (248 nm, 25 ns) (Castillejo et al. 2002 & 2003a) and with a Titanium:Sapphire laser (795 nm, 120 fs) (Gaspard et al. 2008a, b). The colorimetric, spectrofluorimetric and FTRaman measurements provide the basis for discussion of the chemical changes induced on the different tempera paints at the two irradiation wavelengths and pulse durations. As mentioned, enhanced photodegradation of compounds of the egg yolk based binder, mainly dityrosine, DOPA, NFK and kynurenine is observed. However, in the pigmented temperas, the fluorescence bands of the binding medium remain unaltered due to preferential absorption of laser light by the pigment related chromophores, which prevents photo-oxidation of the binding medium components. Irradiation at 248 nm (25 ns) and 795 nm (120 fs) of this type of samples (Castillejo et al. 2002 & 2003a, Gaspard et al. 2008a, b) yielded a similar effect of preservation of the binding medium in presence of the pigment. The reduced extent of changes in the spectroscopic properties of the binding medium in the pigmented samples gives further indication of the stability of the binder in combination with the different studied pigments. These results provide a strong evidence of the prominence of a photochemical mechanism upon laser irradiation at 213 nm (150 ps and 15 ns), with minor contribution of further thermal effects on the surface of the samples, due to the high absorption coefficient of the binding medium at this wavelength (2550 cm−1) (Nevin 2008). Further evidence of the reduced contribution of a thermal mechanism at 213 nm is the absence of carbonization or charring that in contrast does occur at 248 nm, as observed by the appearance of Raman bands characteristics of amorphous carbon (Castillejo et al. 2002). The effect of laser irradiation on the pigmented systems under the conditions of this work is extremely dependent on the composition of the pigment itself. This fact is well illustrated by the comparison of the characteristic spectral features of vermillion and lead chromate in the spectrofluorimetry and FT-Raman results. Strong discoloration in the former, accompanied with the disappearance of the pigment bands, is in contrast with the slight discoloration and unaltered spectral bands in the latter. The limited colour changes (ΔE* ≤ 3.5) observed upon irradiation at 213 nm (150 ps and 15 ns) on the vermillion system are in contrast with those measured upon irradiation at 1064 nm (150 ps), at 248 nm (25 ns) (Castillejo et al. 2002 & 2003a) and 795 nm (120 fs) (Gaspard et al. 2008b).

Figure 3. FT-Raman spectra taken at non-irradiated and laser irradiated areas of tempera paints: a unpigmented and b vermillion. The spectra were taken with 8 cm−1 resolution at the excitation wavelength of 1064 nm.

the position and intensity of the pigment bands, observed between 338 and 403 cm−1 and at 839 cm−1. No bands of the binding medium were observed in the FT-Raman spectra of the pigmented systems, except the corresponding to the aliphatic compounds in the 2750–3100 cm−1 region assigned to the ν(C-H) mode (not shown in the spectra). 4

DISCUSSION

The results presented above obtained under different laser irradiation conditions can be discussed in reference to previous results (Castillejo et al. 2002 & 2003a, Gaspard et al. 2008a, b), although care should be taken when comparing with those obtained on samples prepared with different binding media and under different ageing conditions

20

LACONA8.indb 20

11/10/2010 5:36:10 PM

These differences can be ascribed to differences in the absorption coefficient of the binding medium (2550, 1000 and 100 cm−1 at 213, 248 and 795 nm respectively) (Nevin 2008), resulting in a high absorption of laser photons by the latter at shorter wavelengths. Possible mechanisms of darkening of the vermillion pigment have been the subject of various studies. The detection of traces of black cubic system (α’-HgS) by X-ray diffraction (XRD) upon fs laser irradiation (800 nm, 100 fs), suggested the formation of black meta-cinnabar (Zafiropoulos et al. 2003). On the other hand, XPS measurements performed on samples irradiated with pulses of 1064 nm, 6 ns, have detected a significant increase of the Hg/S ratio which has been attributed to the reduction of HgS to form the darker compound Hg2S (Pouli et al. 2001 & 2003). Hence, the results obtained upon irradiation of vermillion at 1064 nm, 150 ps, could be attributed to the participation of the two mentioned mechanisms. These results also give strong evidence of the little influence of the pulse duration when this pigmented system is irradiated with infrared wavelengths. In contrast, upon irradiation in the UV region, the extent of induced effects is strongly dependent on the absorption coefficient of the binding medium. Upon UV irradiation, and in the presence of strongly absorbing binder, damage to pigments is prevented. On the other hand, the effect of discoloration observed for the lead chromate system at 213 nm (150 ps and 15 ns) and 1064 nm (150 ps) is in agreement with previous observations obtained by irradiation at 248 nm, 25 ns (Castillejo et al. 2002 & 2003a) and at 1064 nm, 6 ns (Gordon Sobbot et al. 2003). This indicates that, even in the presence of an UV absorbing binder, high absorption of laser light by the pigment related chromophore takes place, hence the induced discoloration observed in this pigment. Previous XPS measurements performed on lead chromate showed that 248 nm, 25 ns, laser irradiation induces a noticeable decrease in the CrO42−/Cr3+ ratio (Castillejo et al. 2002 & 2003a) which indicates the formation of chromium (III) oxide (Cr2O3). Oxide formation by reduction of the original lead chromate could explain the colour shift to grey-green as observed by colorimetry. The differences reported concerning the behaviour of the two studied pigmented systems are directly related to differences on the light absorption properties of the pigment chromophore (Castillejo et al. 2002, Johnson et al. 1970, Zhou et al. 2004), HgS semiconductor for vermillion and the CrO42− ion for lead chromate, to their chemical composition and to their sensitivity to oxidation or reduction.

5

CONCLUSION

Various degrees of discoloration and chemical changes were observed upon laser irradiation at 213 nm (pulses of 150 ps and 15 ns) and at 1064 nm (pulses of 150 ps) of different aged and unvarnished tempera paint systems (unpigmented, vermillion and lead chromate). The differences are attributed mostly to differences in the absorption coefficient of the binding medium (2550 cm−1 at 213 nm and negligible at 1064 nm) and less importantly to the pulse duration (150 ps versus 15 ns). The chemical changes induced in the binder are mainly due to photodegradation of the proteins and lipids of egg yolk. It was observed that the extent of chemical change in the binding medium is reduced in the presence of the pigments, due to the effective absorption of the laser pulse energy by the pigment component of the mixture. The vermillion system behaves differently at the two studied wavelengths; the pigment features remain unaffected upon irradiation at 213 nm and disappear by irradiation at 1064 nm. Lead chromate is discoloured at all studied irradiation conditions with more intense modifications induced by IR irradiation (1064 nm). In the systems studied, extra bands attributed to amorphous carbon, indicative of carbonization and charring are absent, in agreement with previous observations upon irradiation with 795 nm, 120 fs pulses, and in contrast with irradiation with 248 nm, 25 ns pulses. More work is in progress on a larger range of pigmented eggyolk based temperas using short laser pulses also in the femtosecond range. In particular, work aiming at studying the laser-pigment interaction, using the raw pigments, will provide a better understanding of the mechanisms involved.

ACKNOWLEDGMENTS Funding from MEC (Projects CTQ2007-60177C02-01/PPQ and CONSOLIDER CSD200700058) and CAM (Programa Geomateriales S2009/Mat-1629) are gratefully acknowledged. One of the authors (MO) acknowledges the CSIC RTPHC for support. For this study, the treated samples were prepared by our collaborator, the late R. Hesterman.

REFERENCES Andreotti, A., Colombini, M.P., Nevin, A., Melessanaki, K., Pouli, P. & Fotakis, C. 2006. Laser Chem. Article ID 39046. Doi:10.1155/2006/39046. Athanassiou, A., Hill, A.E., Fourrier, T., Burgio, L. & Clark, R.J.H. 2000. J. Cult. Herit. 1: S209.

21

LACONA8.indb 21

11/10/2010 5:36:11 PM

Bartoli, L., Pouli, P., Fotakis, C., Siano, S. & Salimbeni, R. 2006. Laser Chem. Article ID 81750. Doi:10.1155/2006/81750. Bell, I.M., Clark, R.J.H. & Gibbs, P.J. 1997. Spectrochim. Acta A 53: 2159. Burgio, L. & Clark, R.J.H. 2001. Spectrochim. Acta A 57: 1491. Burmester, T., Meier, M., Haferkamp, H., Barcikowski, S., Bunte, J. & Ostendorf A. 2005. Proc. 5th Int. Conf. on Lasers in the Conservation of Artworks (LACONA V), Dickmann, K., Fotakis, C., Asmus, J.F. (Eds.), Springer Proceedings in Physics 100: 61. Castillejo, M., Martín, M., Oujja, M., Silva, D., Torres, R., Domingo, C., García-Ramos, J.V. & Sánchez-Cortés, S. 2001. Appl. Spectr. 55, 8: 992. Castillejo, M., Martin, M., Oujja, M., Silva, D., Torres, R., Manousaki, A., Zafiropulos, V., Van den Brink, O.F., Heeren, R.M.A., Teule, R., Silva, A. & Gouveia, H. 2002. Anal. Chem. 74: 4662. Castillejo, M., Martín, M., Oujja, M., Santamaría, J., Silva, D., Torres, R., Manousaki, R., Zafiropulos, V., Van den Brink, O.F., Heeren, R.M.A., Teule, R. & Silva, A. 2003a. J. Cult. Herit. 4: 257S. Castillejo, M., Martín, M., Oujja, M., Rebollar, E., Domingo, C., García-Ramos, J.V. & Sánchez-Cortés, S. 2003b. J. Cult. Herit. 4: 243. Chappé, M., Hildenhagen, J., Dickmann, K. & Bredol, K. 2003. Proc. 4th Int. Conf. on Lasers in the Conservation of Artworks (LACONA IV), J. Cult. Herit. 4: 264S. Cooper, M.I., Fowles, P.S. & Tang, C.C. 2002. Appl. Surf. Sci. 201: 75. Drake, A. & Moore, K. 2004. J. Vib. Spectrosc. 2: 2. Gaspard, S., Oujja, M., Castillejo, M., Moreno, P., Méndez, M., García, A. & Domingo, C. 2008a. Lasers in the Conservation of Artworks, in: Castillejo M, Moreno P, Oujja M, Radvan R, Ruiz J (Eds.), Proceedings of the LACONA VII, Taylor & Francis Group, CRC Press/Balkema, The Netherlands 41. Gaspard, S., Oujja, M., Moreno, P., Méndez, C., García, A., Domingo, C. & Castillejo, C. 2008b. Appl. Surf. Sci. 255: 2675. Gaspard, S., Oujja, M., Abrusci, C., Catalina, F., Lazare, S., Desvergne, J.P. & Castillejo, M. 2008c. J. Photochem. Photobiol. A 193: 187. Gordon Sobott, R.J., Heinze, T., Neumeister, K. & Hildenhagen, J. 2003. J. Cult. Herit. 4: 276S. Johnson, L.W. & McGlynn, S.P. 1970. Chem. Phys. Let. 7: 618. Lakowicz, J.R. 2006. Principles of Fluorescence Spectroscopy, 3nd edn, Springer, New York. Liu, J.M. 1982. Opt. Lett. 7: 196. Mills, J.S. & White, R. 1994. The Organic Chemistry of Museum Objects, 2nd edn, Butterworth Heinemann, Oxford London.

Nevin, A., Cather, S., Anglos, D. & Fotakis, C. 2006a. Anal. Chim. Acta 573–574: 341. Nevin, A. & Anglos, D. 2006b. Laser Chem. ID 82823. Nevin, A., Osticioli, I., Anglos, D., Burnstock, A., Cather, S. & Castellucci, E. 2007. Anal. Chem. 79: 6143. Nevin, A. 2008. PhD. Thesis, Courtauld Institute of Art, University of London, United Kingdom. Osticioli, I., Nevin, A., Anglos, D., Burnstock, A., Cather, S., Becucci, M., Fotakis, C. & Castellucci, E. 2008. J. Raman. Spectrosc. 39: 307. Oujja, M., Pouli, P., Fotakis, C., Domingo, C. & Castillejo, M. 2010. Appl. Spectrosc. In press. Palumbo, G. & Pratesi, R. 2004. Lasers and Current Optical Techniques in Biology, Comprehensive Series in Photochemistry and Photobiology, Royal Society of Chemistry, Cambridge UK. Pouli, P. & Emmony, D.C. 2000. J. Cult. Herit. 1: S181. Pouli, P., Emmony, D.C., Madden, C.E. & Sutherland, I. 2001. Appl. Surf. Sci. 173: 252. Pouli, P., Emmony, D.C., Madden, C.E. & Sutherland, I. 2003., J. Cult. Herit. 4: 271S. Pouli, P., Bounos, G., Georgiou, S. & Fotakis, C. 2007. Proc. 6th Int. Conf. on Lasers in the Conservation of Artworks (LACONA VI), Nimmrichter J, Kautek W, Schreiner M (Eds.), Springer Proceedings in Physics 116. Pouli, P., Paun, I.A., Bounos, G., Georgiou, S. & Fotakis, C. 2008. Appl. Surf. Sci. 254: 6875. Teule, R., Sholten, H., Van den Brink, O.F., Heeren, R.M.A., Zafiropulos, V., Hesterman, R., Castillejo, M., Martín, M., Ullenius, U., Larsson, I., Guerra-Librero, F., Silva, A., Gouveia, H. & Albquerque, M.B. 2003. J. Cult. Herit. 4: 209S. Tonon, C., Duvignacq, C., Teyssedre, G. & Dinguirard, M. 2001. J. Phys. D: Appl. Phys. 34: 124. Vandenabeele, P., Wehling, B., Monees, L., Edwards H, De Reu M. & Van Hooydonk, G. 2000. Anal. Chim. Acta 407: 261. Weeks, C. 1998. Studies in Conservation 43: 101. Wisniewski, M., Sionkowska, A., Kaczmarek, H., Lazare, S., Tokarev, V. & Belin, C. 2007. J. Photochem. Photobiol. A: Chem. 188: 192. Zafiropoulos, V., Balas, C., Manousaki, A., Marakis, Y., Maravelaki-Kalaitzaki, P., Melesanaki, K., Pouli, P., Stratoudaki, Th., Klein, S., Hildenhagen, J., Dickmann, K., Luk’Yanchuk, B.S., Mujat, C. & Dogariu, A. 2003. J. Cult. Herit. 4: 249S. Zhou, G., Lü, M., Gu, F., Wang, S., Xiu, Z. & Cheng, X. 2004. J. Cryst. Grow. 270: 283.

22

LACONA8.indb 22

11/10/2010 5:36:11 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

The role of the substrate in the laser cleaning process: A study on the laser assisted removal of polymeric consolidation materials from various substrates S. Kogou & A. Selimis IESL-FORTH, Vassilika Vouton, Heraklion, Crete, Greece Department of Physics, University of Crete, Heraklion, Crete, Greece

P. Pouli & S. Georgiou IESL-FORTH, Vassilika Vouton, Heraklion, Crete, Greece

C. Fotakis IESL-FORTH, Vassilika Vouton, Heraklion, Crete, Greece Department of Physics, University of Crete, Heraklion, Crete, Greece

ABSTRACT: This study aims to examine and visualize the role of the substrate in the laser cleaning process and consequently in the choice of the optimum irradiation parameters for a successful cleaning application. Towards these objectives, a series of studies were undertaken on model samples specially prepared to simulate a particularly delicate and challenging issue in conservation practice; the removal of aged and degraded polymeric coatings used upon past and/or unsuccessful consolidation treatments on a variety of surfaces. The model samples comprised of substrates of diverse physicochemical properties (wood, aluminum, quartz slides etc.) covered with thin polymeric consolidation films (Paraloid B72). These samples were irradiated with a variety of laser parameters in order to investigate the presence and extent of any side-effects that may be induced to the substrate at the different cleaning conditions. Herein the results obtained upon UV (from 193 nm up to 355 nm) irradiation of naturally dried polymeric films are presented. Particular attention is given in both the absorptivity of Paraloid B72 film at the operative wavelength as well as the role of pulse duration in the ablation process. Under this scheme laser cleaning tests were also comparatively performed with pulse durations ranging from several nanoseconds (ns) down to several hundreds of femtoseconds (fs). 1

INTRODUCTION

Although the response of some of these materials (i.e. dammar and mastic varnishes) to commonly employed laser irradiation protocols (i.e. 248 nm) has been studied in the past (Srinivasan et al. 1989, Georgiou et al. 1998, Fotakis et al. 2006), little attention has been paid to the composition and properties of the underlying surfaces and their role to the cleaning process. Their contribution gets particularly important in the case of cleaning interventions in which the whole film of the polymeric coating must be removed (i.e. altered consolidation treatments). For strongly absorbed wavelengths, efficient coupling of the applied laser energy to the irradiated material enables layer-by-layer material removal with minimal thermal load or damage to the substrate. On the other hand, under moderate or weak absorption conditions a significant amount/ percentage of the laser energy is inevitably diffused into the bulk. This gets particularly important in

The use of polymeric coatings in conservation is mostly associated with protective and consolidation practices aiming to protect delicate surfaces and to strengthen and support fragile structures. Initially coatings were made of natural resinous solutions (i.e. from tree saps), while in the last few decades synthetic polymers were gradually introduced. These synthetic materials have been extensively used on a variety of substrates with different physicochemical properties such as paintings, stonework (Selwitz 1992), metallic objects (Munger 1984), wood (De Witte et al. 1984) etc. Ageing of these materials and their exposure to environmental conditions influence drastically their degradation, while surface phenomena originating from unsuccessful applications may hinder legibility and integrity of the original surface, making thus their removal indispensable.

23

LACONA8.indb 23

11/10/2010 5:36:11 PM

Figure 2. The technical samples employed for the laser tests with Paraloid B72 films on a) quartz slide, b) aluminum plate and c) wooden coupon. The diameter of all samples is 5 cm. Figure 1. Schematic representation of the two absorbing scenarios; a) the laser irradiation is relatively weakly absorbed by the over-layer and b) conditions of strong absorption and/or ultra-short laser pulses.

absorption of the solution by the substrate material, films of variable thickness were acquired (180–220 μm on quartz, 160–200 μm on metal samples and 110–140 μm on the wooden coupons). All samples were subsequently dried in room conditions for at least 48 hours. Initially, irradiation tests were performed on naturally dried samples. Still, in order to address the situation of aged and deteriorated consolidation films these experiments are currently repeated on artificially light-aged samples.

the case of ultra-thin over-layers (i.e. thin varnish or consolidation layers on paintings) with detrimental results into the laser sensitive paint layers. Nevertheless, the use of ultra-short laser pulses (Küper et al. 1987, Bäuerle 2000, Pouli et al. 2008) has been shown to be able to process even nominally transparent materials, while the effective optical penetration depth and consequently any side-effects on the substrate are significantly reduced. In Figure 1 a schematic representation of these scenarios is attempted, in which IO refers to the incident laser intensity while IR is reflected and IA the absorbed intensity. 2

2.2

Laser irradiation parameters

In this study the results of laser irradiation of Paraloid B72 films in the Ultraviolet (UV) region is presented. Additional to the ns irradiation, irradiation using ultra-short pulses (of pico- and femto-second duration) was also performed. Table 1 shows the technical characteristics of the laser systems employed and the parameters of the irradiation series. Irradiation tests were performed in air, on a spot basis using 1, 2 and 10 pulses, while in several cases (i.e. at 193 nm) a sequence of 40 up to 100 pulses was also considered. The employed wavelengths were chosen in order to investigate different absorption scenarios. As it can be seen from Figure 3, which shows the absorbance of the Paraloid B72 used in this study in the UV region (recorded using a Cury 50 UV-Vis spectrometer on an ultra-thin film of about 2.8 ± 0.2 μm), this material absorbs strongly at 193 nm, while its absorption is very weak at 248 nm (more than one order of magnitude) and becomes practically transparent at 355 nm. Table 2 shows the values estimated for the absorption coefficient and optical penetration depth of Paraloid B72, at the studied wavelengths, based on the absorption spectra shown in Figure 3 according to the Beer-Lambert’s law (Pouli et al. 2005). Thus, for the thicknesses of the films studied herein it is expected that a significant amount of the 248 nm and 355 nm irradiation penetrates the Paraloid film and affects the substrate. This may lead to significant and unpleasant side-effects.

EXPERIMENTAL PART

2.1 Test materials To examine the role of the substrate in the laser cleaning process a series of studies on model samples was undertaken. Technical samples consist of polymeric over-layers applied on different substrates (Figure 2). Paraloid B72 (a copolymer of ethyl methacrylate and methyl acrylate) was employed as over-layer material due to its broad use in conservation interventions on Cultural Heritage objects (Hories 2000, Gettens et al. 1966). Moreover its physicochemical properties are well known and studied (Chiantore et al. 2001, Miliani et al. 2001, Favaro et al. 2006). Thin films were prepared by casting solutions of Paraloid B72 in acetone (20% wt) on Quartz slides (Q), Wooden coupons (W) and Aluminum plates (A). Quartz slides do not absorb UV radiation and thus these samples are indicative of the response of the Paraloid B72 material to the various irradiation schemes, allowing thus direct comparison with the other substrates. In all cases the amount of the applied polymeric solution was the same, still due to the differential

24

LACONA8.indb 24

11/10/2010 5:36:11 PM

Table 1. The laser systems and irradiation parameters employed in this study. Wavelength (nm)

Range of operative Pulse fluences duration (J/cm2)

355

10 ns

0.30–2.80 (quartz) 0.15–2.50 (aluminum) 0.06–2.10 (wood)

QS Nd:YAG 355 (EKSPLA, SL312)

150 ps

0.45–1.30 (quartz) 0.10–1.30 (aluminum) 0.15–1.30 (wood)

KrF Excimer (Lambda physik compex 110)

248

25 ns

0.10–3.80 (quartz) 0.10–3.80 (aluminum) 0.10–12.30 (wood)

KrF Excimer dye system (Laser lab Göttingen)

248

500 fs

0.2–2.1 (quartz) 0.2–2.1 (aluminum) 0.2–2.1 (wood)

ArF Excimer (Lambda physik, compex 110)

193

Laser systems QS* Nd:YAG (Spectron, SL805)

25 ns

Table 2. The absorption coefficient and optical penetration depth of Paraloid B72 at the studied wavelengths. Wavelength (nm)

Absorption coefficient (cm−1)

Optical penetration depth (μm)

193 248 355

2600.0 ± 50 115.0 ± 15 4.1 ± 0.2

3.85 ± 0.05 87.00 ± 5 2440.00 ± 100

2.3

Evaluation of the irradiation tests

Initial studies involved a thorough investigation of the laser ablation rates as well as the morphology of the irradiated surfaces. Etching depth rates were measured with a mechanical stylus profilometer (Perthometer) while microscopic observations were performed using an optical microscope (OM) (Nikon, ME 600) employing both transmitted (TL) and reflected (RL) light. 3

RESULTS AND DISCUSSION

Serious damage to the metallic and wooden substrates was observed upon irradiation of the technical samples with the 355 nm (QS Nd:YAG laser, 10 ns pulse duration). Given that this wavelength is not absorbed by the polymeric over-layer (Fig. 3) almost all the laser radiation penetrates the film and interacts with the substrate. This results into severe alterations to the underlying surfaces such as melting, discoloration and surface modifications. The degree and extent of these damages, as well as their threshold values, are closely related to the individual properties of the underlying materials and the pulse duration. For the metallic substrate it was shown that for fluences below 0.6 J/cm2 the laser beam penetrates through the Paraloid B72 film and alters/damages the metal surface. For these fluences no removal of the Paraloid film is observed, still cracking of the polymer was recorded. Figure 4 shows the result of a single pulse of 0.5 J/cm2 in this wavelength, focused on both the metallic and the polymeric film surfaces. It is obvious that under this irradiation condition, no ablation of the polymeric material took place, while serious damage to the metal is visible as discoloration of the aluminum plate. More pulses of the same fluence resulted into polymer material ablation though in an irregular way (craters with uneven and rough edges) and further damage to the substrate. Above 0.6 J/cm2, ablation of the Paraloid film occurs upon single pulse irradiation still the crater morphology appears irregular and with intense

0.25–3.4 (quartz) 0.25–5.3 (aluminum) 0.3–3.9 (wood)

*QS = Q-switched.

Figure 3. The UV-Vis absorption spectra of Paraloid B72 and wood reference (Hon 1991).

25

LACONA8.indb 25

11/10/2010 5:36:11 PM

thermal and/or possibly mechanical alterations (Figure 5a, where the surface around the crater is visibly altered either due to melting and/or stress and strain deformations). It is interesting to note here that the singlepulse ablation threshold for the polymeric film (on quartz) is very high (Fabl = 2.6 J/cm2) while for accumulative pulses (i.e. 10 pulses) the lowest fluence that could result into visible phenomena (bubbling, swelling etc.) was 0.8 J/cm2. The fact that material is removed at lower fluence values in the case of metallic and wooden substrates denotes the presence of secondary effects originating at the interface of the two materials, which are totally uncontrollable. Similar observations were recorded at the ps regime; single-pulse ablation threshold of the neat polymeric film is very high (far beyond the maximum fluence tested, 1.3 J/cm2) while for the metallic and wooden samples damage to the substrate appears, still it is less intense than in the ns regime. Micro-bubble and foaming formation was observed upon irradiation of Paraloid B72 at 248 nm (KrF Excimer laser, 25 ns pulse duration) independently of the substrate (Pouli et al. 2009). In the case of the quartz samples (neat polymeric film) single pulse irradiation resulted into swelling of the polymer for fluences above 0.6 J/cm2, while further pulses were shown to remove material gradually. Figure 6 shows the morphology of areas irradiated in this regime with accumulative number of pulses for 1.95 J/cm2; the 1st pulse causes surface swelling in the range of 20–25 μm while one additional pulse initiates material removal (10 ± 5 μm),

Figure 5. Photo-micrographs of Paraloid B72 film upon 355 nm, 10 ns irradiation using a single pulse of F = 2.8 J/cm2 on (a) the quartz slide and (b) the metal plate (OM- RL, spot diameter 1.5 mm).

Figure 6. Photo-micrographs of Paraloid B72 film on quartz upon 248 nm, 25 ns irradiation at F = 1.95 J/cm2 (a) one, (b) two & (c) ten pulses (OM- TL, spot diameter 1.2 mm). The insets refer to profilometric measurements on the irradiated areas.

which eventually (after 10 pulses) results into a crater of 52 ± 7 μm. Similar phenomena were observed for the other substrates; swelling on the first pulse at all fluences tested (Figure 7) and initiation of etching from the second pulse and onwards. In addition in both cases, no damage occurred on the substrates, as long as they were not exposed to laser irradiation (i.e. due to the smaller optical penetration depth), indicating that the effects of irradiation are restricted in the polymer. Craters with clean edges but structured bases resulted upon 248 nm irradiation in the fs regime (500 fs pulse duration). In contrast to the ns regime the ablated areas show no foaming formation, still nano-holes were evidenced upon SEM observation (Paun 2009). In this regime etching resolution is very fine (in the order of 0.2 ± 0.05 μm) and thus a higher number of pulses is necessary in order to remove the whole thickness of the polymeric film. Still it can be seen that accumulative number of pulses does not cause melting in/or around the crater (Figure 8). Lift-off phenomena were interestingly observed on the sample with the metallic substrate upon multi-pulse (100) irradiation at 2.0 J/cm2 while no damage was detected on the metallic surface. Very fine etching resolution was also found upon irradiation at 193 nm (ArF Excimer laser, 25 ns pulse duration). As this wavelength is strongly absorbed by the polymeric material, it was

Figure 4. Photo-micrographs of Paraloid B72 film on metal upon 355 nm, 10 ns irradiation using (a–b) a single pulse and (c–d) two pulses of F = 0.5 J/cm2. Variable focusing (a–c) on the metal surface and (b–d) on the polymeric film surface reveals the phenomena taking place upon irradiation (OM- RL, spot diameter 1.5 mm).

26

LACONA8.indb 26

11/10/2010 5:36:11 PM

Figure 7. Photo-micrographs of Paraloid B72 film upon single pulse irradiation at 248 nm, 25 ns, F = 3.8 J/cm2 on (a) wood and (b) metal substrates (OM- RL, spot diameter 0.7 mm).

Figure 10. Photo-micrographs of Paraloid B72 film upon 193 nm, 25 ns irradiation on (a) quartz slide at F = 1.25 J/cm2 single pulse (OM- TL, spot area 0.6 × 1.3 mm) and (b) on metal coupon at F = 1.25 J/cm2, 100 pulses (OM- RL, spot area 0.5 × 1.8 mm). The inset refers to profilometric measurements on the irradiated area.

Figure 8. Photo-micrographs of Paraloid B72 film on quartz upon 248 nm, 500 fs irradiation at F = 2.1 J/cm2 (a) single pulse and (b) 100 pulses (OM- RL, spot side length 0.9 mm). Figure 11. Photo-micrographs of Paraloid B72 film on quartz upon 193 nm, 25 ns irradiation (a) 80 pulses at F = 2.5 J/cm2 (OM- TL, spot area 0.5 × 1.8 mm) and (b) 40 pulses at F = 4.5 J/cm2 (OM- TL, spot area 0.4 × 0.7 mm).

with ultra-fine cleaning resolution, in order to remove a layer of rather significant thickness, either a multi-pulse protocol must be followed, or instead, pulses of intense laser fluence should be applied. Both these methodologies may be associated with unpleasant side-effects and thus their choice must be carefully done in order to meet every specific case requirement. Figure 11 shows the phenomena that are associated with multipulse and intense fluence cleaning methodologies. In the first case splashes indicating undesired melting are clearly visible, while, on the other hand, intense fluences may result into cracking and other mechanical alterations. The etching depth per pulse for the 248 nm (both of ns and fs pulse duration) and 193 nm (25 ns) are shown in Figure 12. All depths were determined upon single pulse irradiation except the 248 nm/25 ns case in which single pulse irradiation was shown to result into swelling and bubbling of the surface. In this case the average value of 10 pulses was considered. From the graph it is obvious that upon the 248 nm/25 ns irradiation, etching depth is significantly higher than in the

Figure 9. Photo-micrographs of Paraloid B72 film on metal upon 248 nm, 500 fs irradiation at F = 2.0 J/cm2 (a) single pulse and (b) 100 pulses (OM- TL, spot side length 0.9 mm).

expected that laser radiation would be effectively coupled into the material and “clean” ablation with minimal influence to the surrounding surfaces would be possible. Figure 10 shows the craters obtained upon single pulse irradiation at 193 nm at F = 1.25 J/cm2. In this photo the crater borders and morphology are not easily discernible still their presence is established by profilometric measurements. Actually, the difficulty to differentiate the irradiated area on the polymer surface is the strongest evidence for a clean ablation with sharp crater edges and absence of any thermal or mechanical side-effects both in the crater base and the surrounding area, which is actually the ultimate goal of such an intervention. An important issue raised here is the following: as the strongly absorbed wavelengths are associated

27

LACONA8.indb 27

11/10/2010 5:36:12 PM

REFERENCES Bäuerle, D. Laser Processing and Chemistry (Springer-Verlag: Berlin, 2000). Chiantore, O. & Lazzari, M. 2001. Photo-oxidative stability of paraloid acrylic protective polymers. Polymer 42: 17–27. De Witte, E., Terfve, A. & Vynckier, J. 1984. The consolidation of the waterlogged wood from the GalloRoman boats of Pommeroeul. Studies in Conservation. 29: 77–83. Favaro, M., Mendichi, R., Ossola, F., Russo, U., Simon, S., Tomasin, P. & Vigato, P.A. 2006. Evaluation of polymers for conservation treatments of outdoor exposed stone monuments. Part I: Photo-oxidative weathering, Polymer Degradation and Stability 91: 3083–3096. Fotakis, C., Anglos, D., Zafiropulos, V., Georgiou, S. & Tornari, V. 2006. Lasers in the Preservation of cultural heritage. New York: Taylor and Francis. Georgiou, S., Zafiropulos, V., Anglos, D., Balas, C., Tornari, V. & Fotakis C. 1998. Excimer laser restoration of painted artworks: Procedures, mechanisms and effects. Applied Surface Science 738: 127–129. Gettens, R.J. & Stout, G.L. Painting Materials (Dover, New York, 1966). Hon, D.N.S. Weathering and Photochemistry of Wood (in Hon, D.N.S. & Shiraishi, N. (eds) Wood and cellulosic chemistry, Marcel Dekker, New York 1991). Hories, C.V. Materials for conservation (Butterworth Heinemann, Oxford 2000). Küper, S. & Stuke, M. 1987. Femtosecond UV laser ablation Applied Physics B 44: 199–204. Munger, C.G. Corrosion Prevention by Protective Coatings, (NACE International, Houston, Texas, U.S.A, 1984). Miliani, C., Ombelli, M., Morresi, A. & Romani, A. 2001. Spectroscopic study of acrylic resins in solid matrices, Surface and Coatings Technology 151–152: 276–280. Paun, I.-A., Selimis, A., Bounos, G. & Georgiou, S. 2009. Studies on the UV femtosecond ablation of polymers: Implications for the femtosecond laser cleaning of painted artworks, current volume proceedings. Pouli, P., Melessanaki, K., Giakoumaki, A., Argyropoulos, V. & Anglos, D. 2005. Measuring the thickness of protective coatings on historic metal objects using nanosecond and femtosecond LIBS depth profiling. Spectrochimica Acta Part B 60: 1163–1171. Pouli, P., Paun, I.-A., Bounos, G., Georgiou, S. & Fotakis, C. 2008. The potential of UV femtosecond laser ablation for varnish removal in the restoration of painted works of art. Applied Surface Science 254: 6875–6879. Pouli, P., Nevin, A., Andreotti, A., Colombini, P., Georgiou, S. & Fotakis, C. 2009. Laser assisted removal of synthetic painting conservation materials using UV radiation of ns and fs pulse duration: Morphological studies on model samples, Applied Surface Science 255: 4955–4960. Selwitz, Ch. Epoxy Resins in Stone Conservation (The Getty Conservation Institute, 1992). Srinivasan, R., Braren, B. 1989. Ultraviolet laser ablation of organic polymers, Chemical Reviews 89: 1303–1316.

Figure 12. Etching depth vs. fluence for Paraloid B72 upon irradiation at 193 nm (25 ns) and 248 nm (25 ns and 500 fs).

other two irradiation regimes. On the other hand, in these latter irradiation schemes, which correspond to strong absorption conditions (193 nm/25 ns) and to ultra-short laser pulse ablation (248 nm/500 fs), etching resolution is finer and a plateau of etching depth values is reached. 4

CONCLUSIONS

The role of the substrate in the determination of the cleaning methodology and the laser parameters to be followed upon the laser assisted removal of polymeric over-layers, is herein examined. It is clearly shown that the choice of a laser wavelength which is strongly absorbed by the material to be removed is crucial and in combination with the physicochemical properties of the substrate and the geometry of the over-layer may lead to an efficient cleaning intervention with minimal, if any, damage to the underlying authentic surface. Paraloid B72 is practically transparent to the 3rd harmonic of a Q-Switched Nd:YAG laser at 355 nm and thus a significant amount of the incident laser radiation inevitably interacts with the substrate resulting into detrimental effects. Consequently, cleaning with this wavelength should be avoided particularly in cases in which the underlying surface is laser-sensitive i.e. paintings. The best results were shown to occur upon ns irradiation at 193 nm and fs irradiation at 248 nm. In both cases clean ablation with very high etching resolution (in the order of 0.2 ± 0.05 μm) is achieved and thus both these irradiation schemes fulfil the criteria for a safe cleaning intervention, especially in the case of particularly sensitive and delicate problems such as ultrathin over-layers and highly sensitive substrates. Furthermore, detailed studies are undertaken in order to establish the advantages and limitations of both these two laser ablation regimes, while studies on accelerated aged polymer coatings are also currently performed.

28

LACONA8.indb 28

11/10/2010 5:36:13 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Compact short pulsed fiber laser offers new possibilities for laser cleaning J. Hildenhagen & K. Dickmann Laser Center (LFM), Münster University of Applied Sciences, Steinfurt, Germany

ABSTRACT: Fiber lasers constitute a young but rapid growing category of laser technique. Many advantages such as compact size and no need for maintenance let this technique takes a huge distribution. There by compact short-pulse systems with low output power appear to be interesting for laser cleaning. However, the ablation process takes place by thousands of pulses per second with a laser spot in μm-scale. For applications in the field of cultural heritage especially the possible heat input has to be considered very well. This study was carried out by practical studies and FEM-simulation. It has turned out that the cleaning process easily gains critical temperature and only choosing specific parameter opens the way to clean a limited number of materials. 1

INTRODUCTION

The continuous enhancement of fiber laser sources strengthens their importance in the field of laser materials processing. These laser sources are benefiting from their compact and reliable construction and the cost-effectiveness. However, up to now hardly any efforts have been carried out to apply these laser sources for the cleaning of artworks. First experimental work has demonstrated the potential of cleaning thermally uncritical substrates, e.g. marble, with use of a continuous wave fiber laser source (PL = 300 W) /1/. In addition to continuous wave (cw) fiber laser sources novel pulsed fiber lasers with pulse durations of less than 200 ns and variable pulse shapes (e.g. SPI G3) are available. The low pulse energy (<1 mJ) and pulse repetition rates of more than 10 kHz in combination with a scanner system and focused laser beam to a diameter of less than 100 μm let these laser sources also be applicable for laser cleaning. Among others, this combination was proven by Scholten et al. with a Nd:YAG-laser source /2/. It still has to be clarified which differences are generated in this case by this novel laser beam source. 2

Figure 1. Experimental setup of fiber laser SPI G3 in combination with scanner system.

between the pulses and additionally pulse shape and pulse duration. The possible heat input during the cleaning process built the main focus of this study. Comparisons to a usual short pulse Nd:YAG laser were done during practical cleaning tests which were assisted by pyrometer measurements. In addition, by means of FEM simulations for both laser systems models were created to clarify if the fiber laser could be an alternative to the Nd:YAG laser in relation to the heat generation.

SET-UP AND GOALS OF THE STUDY

At the Laser Centre, Muenster University of Applied Sciences (LFM) a novel laser cleaning system was developed consisting of a compact pulsed fibre laser source (SPI G3, PL = 20 W) and f-theta scanner system (Raylase SuperScan 15), s. Figure 1. The processing results can be affected by variation of pulse energy, pulse repetition rate and overlap

3

PYROMETER MEASUREMENTS

The interaction between beam and surface during the laser cleaning process takes place within only a

29

LACONA8.indb 29

11/10/2010 5:36:13 PM

Further measurements with higher resolution were realized with a faster pyrometer (Kleiber, response time 30 μs). The measuring point of 1–2 mm was centered on the sample area (12 × 12 mm2). During the linear scanning process the in- and decrease of the temperature can be monitored as the laser spot moves through the measuring spot. However, the spot of the hand guided Nd:YAG laser has such a big size that it is permanently inside the measuring area. Therefore the results are hardly comparable but anyway they show a higher temperature level for the fiber laser (s. Fig. 3)

few nanoseconds. Due to the high intensity often a plasma plume is generated above the processing area. Therefore it is hardly possible to measure the real surface temperature precisely. Two different methods were carried out to obtain quantitative results for a comparison. At first small samples (<2 cm2) were imbedded in thermal insulating material and cleaned by laser light for 30 seconds. Thereby for both laser systems parameter were used which cause the same level of purification. Afterwards the surface temperature was measured via IR-thermometer immediately. This procedure was repeated several times to reduce measurement errors. As shown in Figure 2 the heat input of the fiber laser is three or four times higher than of the Nd:YAG laser. This value is nearly equivalent to the factor of output power: 10 W were needed for the fiber laser and 2.8 W for the Nd:YAG laser for obtaining a good cleaning result.

4

FEM-SIMULATION

A close to reality simulation of the heat input is only limited possible because of missing details about parameters like heat transforming and heat entrain. Furthermore precise calculations can be hindered by differences in material properties, inhomogeneities and temperature-dependent behavior. Thus estimations based on practical measuring and literature values have to be used in combination with some simplifications. Thereby the results for both laser systems are linked with a quantitative error but comparable on a qualitative level. As well FEM-simulations were calculated with laser parameters which cause equal cleaning results on practical studies. The virtual measuring spot was placed in the center of the first laser pulse and afterwards the laser beam moves away from that point. Figure 4 shows the accumulation effect of the fiber laser which causes a local limited but high heating effect up to 1800°C. The Nd:YAG laser induces a heating temperature of only up to about 400°C in spite of a higher pulse energy and a much bigger spot size. Thus the intensity of

Figure 2. Measured heating after laser cleaning (duration of cleaning process: 30 seconds; fiber laser 10 W, Nd:YAG laser 2.8 W).

Figure 4. FEM simulation: Surface heating during the first thirty laser pulses, fiber laser (above) and Nd:YAG laser (below).

Figure 3. Temperature measuring with a fast pyrometer during the laser cleaning.

30

LACONA8.indb 30

11/10/2010 5:36:13 PM

6

the radiation on the surface is nearly the same, the reason is particularly the lower repetition rate. During the Nd:YAG laser pluses there is sufficient time to reduce the generated heat via thermal conduction and radiation. For the fiber laser a similar effect can be reached by reduction of the overlap factor, generated via an increasing scanner velocity. This process effects a better dispersal of the heat input. Acceleration from 40 mm up to 500 mm per second will reduce the temperature to 1600°C. Higher speed levels lead to further reduction of the temperature but offer in many cases no sufficient cleaning effect on many different materials.

5

SUMMARY

The combination of compact fiber laser and optical scanner turned out to be an interesting alternative which could create new ideas for laser cleaning processing. However, at current stage of technical development a higher average output energy is needed for laser cleaning in comparison to a Nd:YAG laser. This is probably the result of the significant lower peak power and the compensation due to a repetition rate in kHz-scale. An increased heat input into the material is the side effect which can affect the original surface local bounded. Corresponding pyrometer measurements and FEM simulations affirm the practical results. Another limitation is the critical focus length which must be kept precisely because of the small process slot. This can be realized in a stationary setup on flat samples but not for many objects with cultural heritage background. However, there are promising approaches for mobile systems which generate a line focus via a one-axis scanner and can be hand guided /3/. Furthermore the developments of fiber laser systems step forwards rapidly and more improvements are possible. At present these systems can only be a reasonable addition but no adequate alternative for a Nd:YAG laser system.

PROBLEMS OF THE SCANNING PROCESS

Systematic and often linear scanning routines generate a micro scale grid pattern on various surfaces which can be seen in worst case by the naked eye. This effect is caused by the inhomogeneous energy density inside the beam whereby it is difficult to find a degree of overlap which leads to an equal spreading on the surface. Excessive overlap leads to an over cleaning effect, on the other hand less overlap leads to residues on the surface. Both processes can generate the above mentioned grid pattern. Only those applications are uncritical in cases where a sufficient gap exists between the ablation threshold of the top layer and the modification level of the ground layer. This situation is often given with technical cleaning applications but infrequently with cultural artworks. In general grid pattern can be visual reduced by the use of circular scanning routines. The scanner optic generates a laser beam with a low depth of focus. Even small distance variances between optic and surface lead to a variation expansion of the spot diameter and consequently another energy density. This process can directly affect the cleaning result if there is no sufficient processing latitude. Therefore the cleaning of uneven and 3D objects by a stationary scanner optic is hardly possible.

REFERENCES Hildenhagen, J. & Dickmann, K. Restauro 7, 466–470, Callwaey, Munich 2009. Scholten, H. et al. Laser Cleaning Investigations of Paper Models and Original Objects with Nd:YAG and KrF Laser Systems, LACONA V Proceedings, PP. 11–18, Springer, Heidelberg 2005. Thorsten Naeser: Ein Laser für Neferhotep, Abenteuer Archaeologie 3, S. 72–73, Verlag Spektrum der Wissenschaft, Heidelberg 2006.

31

LACONA8.indb 31

11/10/2010 5:36:14 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Decontaminating pesticide-exposed museum collections J.F. Asmus University of California, San Diego, La Jolla, CA, USA

ABSTRACT: It is occasionally determined that particular museum collections are contaminated with hazardous toxic chemicals. In the case of archaeological items this happens when artifacts have been recovered from agricultural lands that had been treated with pesticides or herbicides. In other instances insecticides have been applied injudiciously within a past museum conservation program. It is sometimes found that in the past historic buildings and structures have been preserved with hazardous (e.g., leadbased) paints. At least thirty years ago it was learned that excimer laser radiation in the ultraviolet portion of the spectrum could be used to destroy and/or remove toxic and hazardous chemicals from surfaces without damaging those surfaces. The investigation reported herein was performed to determine whether ultraviolet radiation from a simple and inexpensive xenon flashlamp is equally effective. Tests were performed with an ultraviolet flashlamp to remove and destroy Malathion on glass plates and painted surfaces. Destruction rates, Malathion fragment species, and surface damage character were all determined. It is concluded that flashlamp radiation may be effective in decontaminating museum artifacts with as few as three flashlamp pulses at a flux of about 3 J/cm2. 1

INTRODUCTION

could harm workers’ health. Many important archaeology sites are found in regions of ancient human habitation with histories of farming (Figure 2). An additional factor may be the aftermath of a “natural” disaster such the 1966 flood in

In recent decades it has been revealed that widely distributed insecticides such as DDT and herbicides such as Agent Orange are persistent and provoke both short-term and long-term toxic effects in living creatures. Thus, museum personnel, archaeologists, conservators, and scholars may unwittingly come into contact with contaminated artifacts in the course of their work (Figure 1). Artifacts may have been accidentally or deliberately contaminated with toxic chemicals that

Figure 1.

Figure 2. region.

Screening soils from a Romanian excavation.

Romanian archaeological site in agricultural

33

LACONA8.indb 33

11/10/2010 5:36:14 PM

Firenze, Italy. The mud and debris that is swept into museums, churches, and monuments consist of sewage, barnyard soils, and cropland silt. Those who are engaged in rescue, cleanup, repair, and conservation are exposed to the entrained toxic chemicals (Figure 3). In these instances it has not been generally recognized that the recovered artifacts need to be decontaminated of toxic residues as well as cleaned and repaired. Figure 4 reveals the extent of pesticide contamination throughout the farmlands of Europe.

2

In bulk form insecticides and other toxic chemicals are usually disposed of through high-temperature incineration. On surfaces decontamination is frequently accomplished through vigorous washing with caustic chemical solutions. Obviously, unique and fragile museum artifacts that have become contaminated with dangerous substances can not be subjected to either of these remedial measures without considerable risk of alteration, damage, or destruction. On the other hand it has long been know that organic molecules can be broken down through exposure to sunlight. This is most often observed as fading or bleaching. Although solar decontamination can be effective (e.g., in largescale oil spills), it is a very slow process. Further, just as solar radiation breaks down toxic chemicals, it decomposes organic materials such as the dyes, fibers and textiles that comprise much of the fabric of many museum collections. The common fading and deterioration of everyday wearing apparel is in part a consequence of this solar effect. With the invention of high-performance ultraviolet (UV) excimer lasers over thirty years ago it became technically feasible to accomplish high-speed photodecontamination (Radziemski, 1981). Laser technology made it possible to apply arbitrarily high UV fluxes to surfaces so that chemical decontamination could be accomplished rapidly. The UV photon is ideally suited for the destruction of toxic molecules. It is slightly larger than the energy holding these molecules together in their lethal form. Imparting this much energy to the chemical dramatically increases its reactivity. It may fall apart, rearrange, or react in order to deal with its energy content. It is thereby denatured. Further, a laser’s parameters (viz., wavelength and pulse length) may be adjusted to minimize or avoid damage or alteration to the surface being exposed to the ultraviolet light. In the decade of the 1980s this procedure was developed to counteract a perceived chemical warfare (CW) threat from countries as well as from terrorists. Military experiments performed in response to the CW threat revealed that pulsed ultraviolet (PUV) energy could rapidly and efficiently destroy chemical nerve agents such as GD (C2H16PO2F), EA 1699, and VX as well as agent simulants such as bis (2-ethylhexyl) hydrogen phosphite, 2-chloroethyl-ethyl sulfide, and diisopropylflourophosphonate (DIMP). As these chemical nerve agents (effective against humans) are chemically similar to pesticides (nerve agents against insects), it is plausible to assume that highpower PUV light may be effective in destroying insecticide residues on the surfaces of museum

Figure 3. Removing the mud and debris that engulfed Firenze during the disastrous flood of 1966.

Figure 4. Europe.

DESTRUCTION OF TOXIC AND HAZARDOUS CHEMICALS

Malathion contamination (dark shading) in

34

LACONA8.indb 34

11/10/2010 5:36:14 PM

objects without causing surface damage (as in the cases of the skin of exposed military personnel or the surfaces of sensitive military ordnance such as communication equipment and computers). 3

LASER MALATHION EXPERIMENT

In order to evaluate the hypothesis that PUV laser irradiation ought to be able to denature insecticides on surfaces a series of irradiation experiments were performed with the ubiquitous pesticide, Malathion, as a test case. For the initial experiments a KrF excimer laser (at a wavelength of 0.25 um) was operated in a single-shot mode and adjusted to a pulse energy of 1J. A UV-grade fused silica lens was used to produce a spot size of 1 square centimeter incident on a glass slide (Figure 5). Analytical grade Malathion was applied to the 1 cm spot on the slide by dissolving 500 ug in CH2Cl2, applying a drop, and allowing the solvent to evaporate before firing the laser. The photodestruction of the Malathion was tracked by measuring (with a UV photodiode and appropriate filters) the fluorescence emitted from the irradiated zone during each laser pulse. The results are shown in Figures 6 and 7.

Figure 7. Malathion decontamination by PUV irradiation.

The approximate data presented in Figure 7 from cursory probative experiments establishes that PUV radiation has comparable effectiveness in Malathion decontamination as has been observed with chemical nerve agents. Elsewhere (Asmus, 1986), it has been proven that for surface cleaning laser fluences and fluxes in this range can be applied to representative museum artifacts without causing damage. Thus, it appears that the application of PUV radiation to the surfaces of contaminated museum artifacts may be a plausible route to their decontamination. 4

DECONTAMINATION WITH XENON FLASHLAMP RADIATION

Although the military decontamination program focused on the use of excimer lasers, this approach presents several impediments to implementation in the museum field. These stem directly from the character of the excimer laser. Specifically, they are large, complex, inefficient, and expensive. Thus, they are an impractical tool to be introduced into a museum or conservation environment except under special circumstances. There is a widely utilized alternative route to the production of high-power PUV radiation. This is the xenon flashlamp that has been the key technology in “flash” and “strobe” photography for more than a half century. Advances in the strength of glass-to-metal seals and the purity of fused-silica (quartz) tubes that transmit UV have led to the development of flashlamps that are efficient and powerful PUV sources. Although decontamination with UV lasers has been shown to be primarily a thermal effect, rather than a photochemical effect, the light radiation does

Figure 5. Malathion spot (smudge near center) on UVgrade fused silica plate.

Figure 6. Malathion fluorescence from the Figure 5 spot during repeated laser shots. The amplitude of shot #4 represents the baseline signal of the bare (decontaminated) glass substrate fluorescence.

35

LACONA8.indb 35

11/10/2010 5:36:14 PM

A test program was formulated to explore the effectiveness of flashlamp PUV for the decontamination of surfaces with pesticide contamination. Accordingly, closed cells (Figure 8) were built out of a short segment of aluminum tube and two 1/16" thick, 3" diameter quartz windows sealed on the ends of the tube by o-rings. Their interiors were cleaned to analytical standards and 500 ug

of analytical grade Malathion was deposited on the inside of the bottom window and the solvent was allowed to dry completely before the cells were sealed. Two cells were irradiated at a flashlamp distance of 6 cm. A first cell was irradiated once and a second one ten times. A third cell, to be used as a control, was not irradiated. Immediately after the irradiation an optical calorimeter was placed at the same location as the cell and the optical energy density was measured as 2.9 J/cm2 over a circular aperture 3.8 cm in diameter centered on the middle of the 5" linear flashlamp (13 mm bore). The two irradiated cells, and the control, were then analyzed by GC/MS. This was accomplished by washing each disassembled component of each cell with CH2Cl2 to detect the Malathion remaining on the bottom window and any that could have redeposited on the walls or cover window. Samples of gas extracted from the cells were also analyzed by GC/MS. It was determined that 15% of the Malathion was removed by one pulse, 10% of that amount was found redeposited on the walls of the cell. After ten pulses 75% of the Malathion was removed from the bottom window, and none was detected on other portions of the cell. The gas from the cell exposed to one pulse contained CO2 (Figure 9), C2H5OH, and t-C4H10. The gas from the cell exposed to ten pulses contained CO2, C2H5OH, i-C3H5CH3. A second test series was performed in order to determine whether a higher flashlamp intensity would increase the rate of Malathion destruction. Unfortunately, operating flashlamps at higher intensities reduces lamp life, which entails more frequent lamp replacement and a correspondingly higher process cost. In addition at higher intensities flashlamp failure sometimes entails quartz envelope explosion leading to hardware and target damage.

Figure 8. Sealed metal cell for flashlamp irradiation of toxic chemicals.

Figure 9. IR spectral scan of the Malathion vapor residue after flashlamp irradiation.

contribute somewhat to decomposition of the vaporized chemicals. In this regard, the polychromatic light produced by flashlamps is advantageous relative to monochromatic light produced by lasers. Monochromatic light tends to break only those chemical bonds that are activated by wavelengths of the particular laser line. The breaking of particular bonds in certain toxic chemicals that may occur when exposed to monochromatic (laser) light, may produce daughter chemicals that are also toxic. Broadband radiation (e.g., flashlamp light), on the other hand, would be expected to photochemically decompose a wider range of chemical bonds and to fragment the contaminant into more elementary units, which are less likely to be toxic. In summary it appears that PUV from xenon flashlamps may be both more cost effective and more thorough than excimer lasers in decontaminating surfaces. The only advantage accompanying the use of a laser would be its ability to send a highly collimated beam over great distances. This, of course, is not relevant in a museum environment. Thus, experiments were performed to determine the prognosis for flashlamp insecticide decontamination for application in the museum and/or conservation laboratory. 5

XENON FLASHLAMP FEASIBILITY EXPERIMENT WITH MALATHION

36

LACONA8.indb 36

11/10/2010 5:36:15 PM

applied to the surface as illustrated in Figure 5. Next, a nebulizer was employed to puff a cloud of Nile Blue vapor onto each spot of Malathion. The reference dyed Malathion film spot appears at the upper right in Figure 10. The aftermath of the flashlamp treated spot is just to the left of center in the figure. There is no evidence of residual dye. The Malathion and Nile Blue have been incinerated to a carbon and hydrocarbon ash that is visible on the left side of Figure 10. We found that this ash could be either wiped or rinsed away quite readily. A demonstration of flashlamp decontamination of a Ming Dynasty stone (Dolomitic) statue is shown in Figure 11.

An alternative approach was instituted in order to enhance the effectiveness of decontamination by means of high-power pulsed flashlamp illumination. It was reasoned that contaminated surfaces could be pretreated by first spraying with a harmless dye such as Nile Blue. First, this would increase the absorption of the light by the virtually transparent Malathion film. Consequently, the xenon lamp could be pulsed at a lower energy, thereby increasing lamp life and reducing process cost. Second, the resultant bleaching and/or incineration of the dye would be a visual indication of the thoroughness of the surface coverage. Finally, the degree of bleaching of the dye is a much more convenient monitor of the degree of decontamination than having to resort of a more complex diagnostic technique such as surface spectral fluorescence. Figure 10 shows an example of the result of dye-enhanced flashlamp decontamination of a glass test plate. Initially, two Malathion drops were

6

CONCLUSION

These probative experiments have shown efficient removal of Malathion as well as its virtually complete denaturization by means of UV radiation from an excimer laser. In addition it has been demonstrated that conventional xenon flashlamp light is able to destroy Malathion on non-absorbing surfaces. However, it is clear that flashlamps produce sufficient flux only at an excitation approaching the explosion limit. Consequently it is concluded that if this decontamination process is to be employed, an absorbing dye should be applied to the surface prior to irradiation in order to improve optical absorption and enhance the interaction. The preliminary work reported here needs to be extended to tests on absorbing surfaces and actual historical artifacts. (Radziemski, 1981). REFERENCES

Figure 10. Malathion decontamination by flashlamp irradiation (absorption enhanced by application of NileBlue dye).

Asmus, J. 1987. More light for art conservation. IEEE Circuits and Devices (March): 6–15. Radziemski, L. 1981. Laser-induced photodestruction of the organo-phosphates: DIMP and DMMP. J. Environ. Sci. Health B16 (3) 337–361.

Figure 11. Flashlamp decontamination of the stone of a Ming-Dynasty statue of a spirit path guardian.

37

LACONA8.indb 37

11/10/2010 5:36:15 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Laser cleaning of burial encrustation and aged protective coating on Egyptian leather: Optimization of working conditions A.A. Elnaggar Conservation Department, Faculty of Archaeology, Fayoum University, Alfayoum, Egypt

P. Pouli Institute of Electronic Structure and Laser, FORTH, Heraklion, Crete, Greece

A. Nevin Courtauld Institute of Art, London, UK

M.A. Fouad Conservation Department, Faculty of Archaeology, Cairo University, Giza, Egypt

G.A. Mahgoub Restoration Sector, The Supreme Council of Antiquities, Cairo, Egypt

ABSTRACT: Egyptian leather objects from between 1050–1500 BC from Luxor, (Egypt) have been selected to assess the potential of using laser-assisted cleaning for the removal of burial encrustation and aged protective coatings using a 20 nanosecond Nd:YAG operating at 1064 nm. Investigations were motivated by the extreme difficulty and risk associated with mechanical cleaning of fragile materials and the need for an alternative method. First, damage threshold fluences were determined when operating at the employed wavelength with fresh, artificially aged and real samples of ancient vegetable tanned leather. A series of analytical techniques including X-ray diffraction, optical and scanning electronic microscopy, colorimetery and Fourier Transform Infrared Spectroscopy have been used to assess the extent to which laser-assisted removal of material induces morphological and chemical changes, to evaluate the laser cleaning results compared with mechanical cleaning method, and to determine optimum laser cleaning parameters. 1

INTRODUCTION

laser for cleaning in most applications is the strong absorption of most unwanted layers (i.e. pollution encrustation etc) to this wavelength in contrast to the substrate (Larson et al. 2000). Currently, Laser cleaning is often considered to be an alternative and practical technique offering the conservator a high level of precision and control. However, leather artefacts have received very little attention, and the majority of the reported laser cleaning work on collagen objects has been focused on the cleaning of parchment with using Infrared, visible and ultraviolet pulsed lasers which successfully removed soot and dirt particles without damaging the substrate fibres, and on the other hand, unacceptable damage was reported in other cleaning trails. (Kautek et al. 1995a, Kautang et al. 1995b, Kautek et al. 1998, Kautek et al. 2000, Kautek et al. 2003, Kautek & Pentzien 2003, Sportun et al. 2000, Puchinger et al. 2003, Vest et al. 2003, Kenned, et al. 2004, Batishche et al. 2005).

The traditional methods for the conservation and consolidation of leather have usually involved the application of adhesives or solvents over the surface, following the cleaning of the dirt from the surface using mechanical or chemical methods. The embrittelment of weathered leather artefacts prevents the employment of mechanical methods for cleaning, such as brushing and plastic erasers. Laser cleaning is an alternative little-explored method for such an intervention. The main advantage of laser-assisted removal of material may be that lasers do not interfere mechanically with the collagen fibers of the surface and thus swelling (due to solvent cleaning) or abrasion and tearing (due to mechanical action) are avoided. Among the most commonly used lasers which have been proposed for the removal of dirt from leather is the Nd:YAG laser which is emits in the near infrared at 1064 nm. The success of this specific

39

LACONA8.indb 39

11/10/2010 5:36:16 PM

the thick burial crust found on excavated leather and to test its ability to reverse thin aged protective coatings (Paraloid B72) from model leather samples. Although paraloid is common in leather consolidation (Chiantore & Lazzari 2001, Maria 1996, Abdel-Maksoud 2000) its removal is not straightforward and thus laser cleaning was also assessed. Furthermore this research aimed to assess whether fiber damage will occur with laser-assisted cleaning of samples, and to determine the most suitable cleaning parameters based on a series of tests. Further, as part of the work, analysis of other potential alterations which may take place at the surface of leather as a direct result of irradiation has been carried out. Damage thresholds of fresh and artificially aged leather as well as removal threshold for the crust and the aged protective coating at 1064 nm have been established. The effectiveness of laser cleaning has been tested using optical microscopes and SEM, Colorimetric measurements, XRD and FTIR. Finally, the laser in cleaning process has been tested on real samples and the results have been compared with mechanical cleaning methods.

Leather, in contrast to parchment, is the product of the tanning process on animal skins which occurs with the treatment of hides with vegetable tannin extracts or by chromium ions, used in mineral tanning. In Ancient Egypt, the vegetable tanning (hydrolysable tannins (pyrogallols) was employed, and is based on the treatment of animal hides using natural sources of tannins such as oak wood, gulls and sumac leaves, as well as condensed tannins (catechols) found in acacia, mimosa and pine. Vegetable tannins are large molecules, which interacts and transforms animal leather proteins (collagen) into resistant insoluble products which are chemically stable. These products may have different chemical structures, but it has some common properties such as they are insoluble in organic liquids, miscible with water, hygroscopic, amorphous substances, very sensitive to oxidation and reduction (Puică et al. 2006). Collagen-based artifacts include leather, parchment and vellum. Although the total manufacturing process of the three types differs, some manufacture steps are the same. The chemical and physical properties of leather vary significantly from those of parchment; indeed the stability of leather is increased due to the bonding of the fiber network taking place during the tanning process. Many Egyptian leather products have often been preserved intact, thus allowing a much better knowledge of the manufacturing techniques and stages, such as salting, tanning, lubrication and finishing. Deterioration factors of leather may be chemical, physical and biological factors (Dirksen 1993, Larsen & Chahine 2000). Leather that has been exposed to excessive dryness will be cracked, broken, and embrittled. High humidity levels will encourage the growth of mold that will lead to staining, surface distortion, and softening of the leather. Contact with insects may be demonstrated by the presence of small holes and loosened parts. The accumulation of dust can be a serious problem for leather conservation since dust particles can be difficult to remove from the irregular surface of leather. The chemical deterioration of leather is due to acid hydrolysis which may be produced by air pollution or improper storage, and oxidation which may increase with exposure to light: acid hydrolysis is the most aggressive and may lead to faster deterioration. Another major factor of damage associated with leather artifacts is improper cleaning (mechanical and solvent) which may lead to irreversible damage, as well as improper irreversible consolidation and surface treatments which may deteriorate with age, become sticky and therefore seriously compromise the appearance and stability of the objects. The aim of this research was to evaluate the potential of Nd:YAG laser (1064 nm) to remove

2 2.1

MATERIALS AND METHODS Samples

Real samples were selected from vegetable tanned unknown piece of leather dating from pharaonic period (new kingdom 1570–1070 BC, DearEl-Madina, Luxor, Egypt conserved at the Agriculture museum in Cairo under number 5261). Their condition is very bad. The object is fragile, brittle, weaved, and layers of the dirt were strongly adhered to the substrate. Further, small and large cracks have permeated the material, which has significant missing parts and large tears along the

Figure 1. Burial dirt on leather: A piece of vegetable tanned leather dating from the pharaonic period (new kingdom 1570–1070 B.C.E., excavated at Dear- ElMadina, Luxor, Egypt conserved at the Agriculture museum in Cairo (under number 5261).

40

LACONA8.indb 40

11/10/2010 5:36:16 PM

edges. As part of research, new vegetable tanned (mimosa) leather from cattle (6 months old, Special Handmade in Egypt) was used to prepare the test samples used for the preliminary stages of this study, while special attention was paid so that this new leather to be durable and in close similarity and manufacturing technology to the archaeological leather samples (Meghea 2004). The test samples were cut from selected areas of the whole skin to obtain samples with maximum similarity in the direction of the collagen fiber bundles (Hansen & Lee 1992). Tests were also performed on model samples of leather coated with Paraloid B72 (6%, acrylic polymer, copolymer of methyl acryl ate & ethyl methacrylate and butyl methacrylate) diluted in acetone. 2.2

calculated by dividing the pulse energy by the surface of interaction (spot size was estimated by means of a photosensitive PVC film) and are therefore an average value. Damage (Fdam) and ablation (Fabl) threshold measurements To ensure a safe laser-assisted crust removal from leather substrates it was first necessary to establish the damage threshold (Fdam) of the substrate at the employed wavelength and pulse duration. As damage threshold is defined as the maximum fluence values that, upon irradiation, induce no changes to the leather substrates (for the new fresh, aged and ancient samples). Similarly, the ablation threshold values (Fabl) were established as the minimum fluences to initiate ablation of the over-layer (Paraloid B-72 and burial crust). 2.3.1

Artificial thermal ageing

Artificial ageing is an absolute necessity for testing the effect of conservation treatments and the durability of new leathers, and to imitate an average deterioration observed for historical leathers. Heat aged leather may be obtained within a few days using dry heat at temperature above 100°C, and under 120°C to avoid the drastic physical changes which are documented in the deterioration of leather. The artificial ageing treatments were performed on new vegetable (mimosa) tanned leather. The leather has been submitted to the thermooxidative aging in standard laboratory oven with dry heat at 100°C for 11 days (Larsen 1993, Larsen 2000). The changes induced during aging have been monitored daily and was accompanied by a decreasing of thickness and overall dimensions, decrease in mass and various morphological changes. Aged leather coated with Paraloid B72 has been submitted to photo-oxidative ageing in an ageing chamber (custom-made at IESL-FORTH) for 480 hours of exposure to daylight irradiation at 35°C with Graphica PRO TLD Philips Lamps (Intensity: 10000 Lmm−2) (Arbizzani et al. 2004). The ageing of Paraloid has been carried out to produce samples which are similar to naturally aged leather treated with consolidants to assess laser—cleaning. 2.3

3

ANALYTICAL INVESTIGATIONS

CIE*Lab System, (Sheen Micromatch plus, Gloss Zoom 20 mm, Sheen Instruments Ltd., England) Colorimeter served to measure the colour changes of the different leather conditions (new, aged, coated with polymer and irradiated by laser) based on changes observed on the L scale (Luminosity), a* scale (red/green) and b* scale (yellow/blue). Five measurements were averaged to obtain one data point. X-ray diffraction (XRD) analysis was used to determine the chemical composition of the burial encrustation on real samples, and was carried out using RIGAKU, RINT 2000 Series. Optical and scanning electron microscopes have been used to determine the damage and ablation threshold fluences of leather based on observations of the morphology of treated surfaces and to observe the burial crust and aged polymer removal. Fourier Transform Infrared Spectroscopy (FTIR) was carried out using a Thermo Nicolet iS10 FT-IR Spectrometer in reflectance mode using an ATR (Attenuated Total Reflectance) slide-on accessory with diamond crystal, spectral range 4000–400 cm−1 and resolution 4 cm−1, to determine any chemical structure changes in model and real samples before and after laser cleaning, and to compare vibration spectra of mechanically and laser—cleaned samples.

Lasers

Laser experiments were performed using Q-switched Nd:YAG laser wavelength (Spectron SL805) emitting at 1064 nm with pulse duration in the range of 20 ns. The beam was delivered by an articulated arm to clean the samples surfaces which involved moving the samples in micrometer steps which was carried out using a manual stage. The laser was operated mainly in a singleshot mode and fluences were regulated between 1.85 J/cm2 to 7.5 J/cm2 per pulse. Fluences were

4

RESULTS AND DISCUTION

4.1 XRD analysis Molecular analysis using XRD indicated that the burial encrustation of real leather consists predominantly of Sodium Aluminum Silicate (NaAlSi3O3).

41

LACONA8.indb 41

11/10/2010 5:36:16 PM

remove the whole layer, because the polymer has been penetrated among the leather fibers. Removal of the burial encrustation on real leather could be performed safely at the ablation threshold fluence (3.2 J/cm2).

This indicates that the black crust is silicaceous dirt and clay which was attached to the leather surface before excavation. 4.2

Colorimetery

From colorimetric measurements shown in Table (1), it is evident that the model samples were severely darkened after the aging and the polymer application while the aging of the Paraloid-covered sample and the laser cleaning tests did not cause significant changes. Specifically, the red-green value (a*) of the sample with noticeably increases upon aging of the fresh sample and following the application of consolidant, while on the contrary the yellow-green value (b*) was found to change after aging of the polymer-coated samples and laser tests, indicating evident yellowing of the surface. 4.3

4.4

FTIR analysis

FTIR analysis of fresh and aged leather (Fig. 2) shows that standard collagen peaks are present in the spectra. Slightly better defined peaks including a small band at 1377 cm−1, which is more evident in aged leather than in non-aged leather, are seen. This new peak could be related to the accumulation of an unidentified degradation product in the leather sample. FTIR-ATR spectra of Paraloid B72 over leather (Fig. 3) contain bands from collagen-based leather

Determination of damage (Fdam) and ablation (Fabl) threshold fluences

0.12

Aged Fresh

0.11 0.10

A series of laser irradiation tests, optical and scanning electron microscopes investigations were carried out to determine the damage and ablation threshold fluences at1064 nm which are shown in Table (2). It is clear that the ablation threshold fluence (4.7 J/cm2) of aged polymer (Paraloid B72) on fresh and aged leather is higher than the damage threshold fluence of fresh and aged leather. Therefore, it is evident that it is not possible to remove this polymer using laser radiation at 1064 nm without damaging the leather substrate. Laser cleaning of aged polymer at 4.7 J/cm2 was successful to reduce the polymer thickness but was not able to

0.09

Absorbance

0.08

a*

b*

Reference Upon aging of fresh leather After Paraloid B-72 application After aging of Paraloid B-72 After laser irradiation tests

78.79 52.73 39.55 38.38 37.78

7.69 18.83 24.06 26.10 26.65

15.83 17.33 21.02 28.48 28.99

0.05

0.03 0.02 0.01 0.00 3500

3000

2500

1800

1600

1400

wavenumber/cm

1200

1000

800

−1

Figure 2. FTIR spectra of fresh and aged leather, with peaks associated with fresh leather are indicated with an arrow. Polymer over aged leather Laser cleaned polymer over aged leather Aged leather *

Absorbance/arbit. units

L

0.06

0.04

Table 1. Colorimetric measurements of model leather samples with different conditions. Sample

0.07

Table 2. Damage (Fdam) and ablation (Fabl) threshold fluences of leather in different conditions.

*

*

1064 nm Leather condition

Fdam (J/cm2) Fabl (J/cm2)

3500

3000

2500

1800

1600

1400

1200

1000

800

−1

wavenumber/cm

Fresh leather 3.9 Aged leather 4.5 Aged polymer on aged leather Burial crust on real leather

Figure 3. FTIR spectra of aged leather, aged polymer over aged leather and leather surface after laser cleaning of aged polymer (at 4.7 J/cm−2). Peaks which differ are indicated with an asterisk (*).

4.7 3.2

42

LACONA8.indb 42

11/10/2010 5:36:16 PM

as well as characteristic bands from Paraloid B72. Carbonyl C-O stretching at 1728 cm−1 is the strongest band from the acrylic polymer; other bands at 1385 cm−1 and 1150 cm−1 are also ascribed to the presence of the polymer. In the C-H stretching region, the shoulder at 2980 cm−1 is specifically associated with the Paraloid B72. Laser cleaning at the ablation threshold of the aged polymer from the leather has not succeeded in removing the entire polymer as can be appreciated from a comparison of FTIR spectra of the cleaned leather and the aged leather without any polymer. After cleaning, peaks at 1728 cm−1, 1150 cm−1 and 1385 cm−1 are significantly reduced in comparison with the Amide-I stretching vibration at 1634 cm−1 in leather. Using fluence higher than ablation threshold fluences to remove the whole polymer film will damage the leather substrate. Therefore, some of the polymer is still remains on the surface. No change in the peaks ascribed to leather is seen. FTIR spectra of surface dirt is characterised by peaks ascribed to gypsum and other sulphates with a broad peak centred at 997 cm−1, a shoulder from silicates at 1080 cm−1 and bands from calcite at 874 cm−1 and 1425 cm−1, and this is confirming the XRD interpretation of burial crust. Further, signal from the substrate Amide-I and Amide-II are appreciable. FTIR-ATR spectra comparing mechanical and laser cleaning of dirt (Fig. 4) from a sample of leather highlight negligible differences between samples. Both samples contain spectra typical of leather (collagen) with Amide-I (1634 cm−1) and Amide-II (1534 cm−1) bands dominant. Only the peak ratios of bands at 1030 cm−1 and 1080 cm−1 are slightly different in laser cleaning than mechanically cleaned samples. Peaks at about 1080 cm−1 are found in L-proline, trans-4- hydroxy-proline and gelatine and are related to the C-O stretching or skeletal stretching.

5

CLEANING PROCEDURES

After optimization of the laser cleaning fluences and pulses number at 1064 nm, satisfactory laser cleaning at fluence 3.2 J/cm2 and 5 pulses was carried out to remove the burial crust from real leather surface. Mechanical cleaning with a scalpel and brushing of the burial crust gave unsatisfactory results because of damaging the underlying collagen fibers.

Figure 5. SEM of burial encrustation on leather surface before cleaning.

Figure 6. SEM of Archaeological leather after laser cleaning at 3.2 J/cm2.

Laser cleaning of dirt Mechanical cleaning of dirt Surface dirt

0.05

Absorbance

0.04

0.03

0.02

0.01

0.00 3500

3000

2500

1800

1600

wavenumber/cm

1400

1200

1000

800

−1

Figure 4. FTIR spectra of surface dirt before cleaning, leather surface after laser and mechanical cleaning.

Figure 7. SEM of Archaeological leather after mechanical cleaning.

43

LACONA8.indb 43

11/10/2010 5:36:16 PM

Arbizzani, R. et al. 2004. Decay markers for the preventative conservation and maintenance of paintings. Journal of cultural heritage, 5, Issue 2, April— June, 167–182. Batishche, S. et al. 2005. Simultaneous UV-IR Nd:YAG laser cleaning of leather artifacts. Lasers in the Conservation of Artworks: LACONA VI Proceedings, Vienna, Austria, September, 21–25. Chiantore, O. & Lazzari, M. 2001. Photo-oxidative stability of paraloid acrylic protective polymers, Polymer, 42, Issue 1, January, 17–27. Dirksen, V. 1993. The degradation and conservation of leather. Journal of conservation & museum studies, 3, November. Hansen, E.F. & Lee, S.L. 1992. The Effects of Relative Humidity on Some Physical Properties of Modern Vellum: Implications for the Optimum Relative Humidity for the Display and Storage of Parchment’, JAIC, 31, Number 3, Article 5, 325–342. Kautek, W. et al. 1995a. Laser cleaning of ancient parchments. Lacona I, Laser in the Conservation of Artworks, 4–6 October, Heraklion, Crete, Greece, 69–78. Kautek, W. et al. 1995b. Probing the limits of paper and parchment laser cleaning by multi-spectral imaging. Lacona I, Laser in the Conservation of Artworks, 4–6 October, Heraklion, Crete, Greece. Kautek, W. et al. 1998. Laser interaction with coated collagen and cellulose fibre composites: fundamentals of laser cleaning of ancient parchment manuscripts and paper, Applied Surface Science, 127–129, May, 746–754. Kautek, W. et al. 2000. Near-UV laser interaction with contaminants and pigments on parchment: laser cleaning diagnosis by SE-microscopy, VIS, and IR spectroscopy, Journal of cultural Heritage 1, S233–S240. Kautek, W. et al. 2003. Diagnostics of parchment laser cleaning in the near-ultraviolet and near-infrared wavelength range: a systematic scanning electron microscopy study. Journal of Cultural Heritage. 4, Supplement 1, January, 179–184. Kautek, W. & Pentzien, S. 2003. Laser cleaning system for automated paper and parchment cleaning. In K. Dickmann, C. Fotakis & J.F. Asmus (eds) Lasers in the Conservation of Artworks, LACONA V Proceedings, Osnabrück, Germany, September 15–18. Kenned, C. et al. 2004. Laser cleaning of parchment: structural, thermal and biochemical studies into the effect of wavelength and fluence. Applied surface science 227, 151–163. Larsen, R. 1993. Step Leather Project. Evaluation of the correlation leather and determination of parameters for standardization of an artificial aging method, Second Progress Report, March, 229. Larsen, R. 2000. Experiments and observations in the study of environmental impact on historical vegetable tanned leathers, Thermochimica Acta, 365, No. 1–2, 85–99. Larsen, R. & Chahine, C. 2000. The deterioration and conservation of vegetable tanned leather in ancient books. In: Grau, A.B & Vitoria, S.P (Eds), Research for protection, conservation and enhancement of cultural heritage: opportunities for European enterprises 4th European Commission Conference on Strasbourg (FR), 22–24 November. Larson, J., Cooper, M. & Sportum, S. 2000. Developments in the application of laser technology for

Figure 8. Successful laser cleaning of dirt on archaeological leather surface (left) and an uncleaned area (right).

6

CONCLUSION

After optimization of laser cleaning parameters and the analysis of the burial crust, determination of damage threshold of aged model leather and the ablation threshold of the burial crust over the archaeological leather was assessed. Successful cleaning of the burial crust over archaeological leather has been carried out with 1064 nm operation at about 20 ns with fluence no greater than 3.2 J/cm2. The burial crust was removed with 5 pulses. The leather cleaning was assessed using optical and scanning electron microscopy. The results demonstrated that the laser cleaning of the burial crust was more successful and did not cause damage of fibres structure, unlike the results from mechanical cleaning. Paraloid B72 in highly used in Egypt as a consolidant for similar leather objects. For optimum removal of the aged film of Paraloid B72 over aged model leather, series of laser cleaning tests, colorimetric measurements and FTIR analysis have been carried out. The result demonstrated that the laser cleaning of thin aged film of Paraglide B72 over aged model leather at 1064 nm using the ablation threshold fluence at 4.7 J/cm2 was not successful to completely remove the aged polymer without damaging the leather fibres because the ablation threshold fluence of aged polymer is higher than the damage threshold fluence of aged leather. Further work will concern the study of mechanical properties of leather after laser and mechanical cleaning, and the effect of laser cleaning on different tanning materials. REFERENCES Abdel-Maksoud, G. 2000. An evaluation of selected applied polymers for treatment of parchment. 15th World Conference on Nondestructive Testing, Roma, 15–21 October.

44

LACONA8.indb 44

11/10/2010 5:36:17 PM

conservation. In Roy, A. & Smith, P (Eds), Tradition and innovation advances in conservation, Contributions the Melbourne congress, 10–14 October, London. Maria, M. 1996. Preservation and conservation of archaeological objects in leather and vegetable fibres. Preprints of the contributions to the Copenhagen congress, 26–30 August, 27–31. Meghea, A. 2004. Behavior to Accelerate Ageing of Some Natural Biopolymer Constituents of Parchment. Mol. Cryst. Liq. Cryst., 418, 285/(1013)–290/(1018). Puchinger, L. et al. 2003. Chemistry of parchment-laser interaction. In K. Dickmann, C. Fotakis & J.F. Asmus (eds) Lasers in the Conservation of Artworks, LACONA V Proceedings, Osnabrück, Germany, September 15–18, 51.

Puică, N.M. & et al. 2006. FTIR spectroscopy for the analysis of vegetable tanned ancient leather. European Journal of Science and Theology, December, 2, No. 4, 49–53. Sportun, S. et al. 2000. An investigation into the effect of wavelength in the laser cleaning of parchment. Journal of Cultural Heritage, 1, Supplement 1, S225-S232. Vest, M. et al. 2003. Evaluation of laser cleaning of parchment documents with a Q-switched Nd:YAG laser at 1064, 532 and 266 nm. In K. Dickmann, C. Fotakis & J.F. Asmus (eds) Lasers in the Conservation of Artworks, LACONA V Proceedings, Osnabrück, Germany, September 15–18, 217–225.

45

LACONA8.indb 45

11/10/2010 5:36:18 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

The practical use of lasers in removing deteriorated Incralac coatings from large bronze monuments A. Dajnowski Director of CSOS Inc, Illinois, USA

A. Lins Chairman of Conservation Department, Philadelphia Museum of Art, USA

ABSTRACT: This paper discusses the use of laser ablation for the practical removal of a deteriorated Incralac coating on the monumental bronze sculpture of Tadeusz Kościuszko by Kazimierz Chodziński, which has been exposed outdoors in Chicago since its installation in 1904. Current in situ cleaning options in the conservation field—primarily poulticing, chemical stripping, or abrasion—present environmental problems, especially containment and disposal of chemical runoff or abrasives which can prove very difficult and expensive to resolve. Laser technology provides conservators with a cost effective and safe tool for the controlled removal of failed coatings, avoiding environmental issues and their financial consequences. The determination of optimal laser conditions for coating removal are described for this project, together with an evaluation of the cause of corrosion under very deteriorated areas of the failed coating, the properties of the failed coating and its ablated by products. The form and size of the ablated organic coating particles as well as the inorganic debris are illustrated, together with energy dispersive spectra. Preliminary tests to evaluate how differential absorption by different corrosion phases, states of aggregation and hydration affected the lacquer removal are discussed. 1

INTRODUCTION

become tunnelled by iron corrosion and blanched in many places by absorbed water. The laser treatment in 2008—described below—focused on coating removal and removal of the continously problematic chlorides. The treatment took approximately one month.

The Kościuszko Monument has been moved three times since its installation in 1904. The latest move was in 2008 as part of a major reconstructuion of Burnham Park which involved moving the monument approximately 50’ east. The bronze monument (approximately 83Cu: 10Sn:4Zn;3Pb) had been vandalized repeatedly and suffered decades of corrosion and lack of maintenance when the monument was first treated in 1997. The protective Incralac coating—applied in the 1997 treatment—had begun to fail in 2003. The surface was chemically cleaned of failed coating using solvent based poultices and steam, then repatinated with a mixture of ferric nitrate and potassium permanganate to approximate the original color. It was retreated with BTA and coated with Incralac. By 2008 signs of deterioration and corrosion were again evident, caused by ongoing chloride corrosion. BTA—applied in 1997 and 2003—had failed to stop active chloride pitting and its undermining of the protective coating. The coating had

Figure 1. Detail from laser cleaning of Kościuszko Monument—2008.

47

LACONA8.indb 47

11/10/2010 5:36:18 PM

Figure 2. 2009.

SEM. Samples 12, 14,16, 20 and 21 were analyzed without duplication. All samples were examined under a high vacuum at 20 kV; EDS spectra were collected with an Oxford INCA X-sight detector. 2. Preliminary laboratory tests: Three test surfaces were prepared: Powders of Cu (Johnson Matthey, 99.99%), Cu2O (Aesar 99.5%), brochantite [1], copper acetate monohydrate (Fisher, 98%), anhydrous cuprous chloride (Fisher, purified), anhydrous cupric chloride (Fisher, 96.8%) were set into 4 mm diameter holes × 3 mm deep in: a) a plexiglas© block (PMM), and b) a 90Cu10Sn bronze disk. The powders were compacted and scraped level with the surface. A thin coating of Incralac (15%w/v) was applied over each hole. A 90Cu10Sn cast disk was polished to 3 micrometers and then etched heavily over one half of the surface with a solution of FeCl3 (10 g) and HCl (20 ml) in 100 ml H2O and lightly etched with the same etchant over the other half. Two strips of continuous cast 90/10 bronze were prepared, one with predominately brochantite on the surface, one with predominately cuprite on the surface. Each was coated with Incralac (as above) in three different thicknesses. Each sample was exposed to the beam of an Adapt 20W Backpack laser (equipped with 250 mm lens) for 10–15 seconds at an estimated fluence of 1.9 J/cm2.

Kościuszko Monument after treatment—

2.1

The first step of the treatment consisted of tests performed to establish the appropriate fluence of the beam and to safely remove the deteriorated coating. The fluence shown to effectively remove the coating without removal of stable corrosion products was in the range between 3 and 4 J/cm2 at a frequency of impulses set to 30 KHz, scan frequency 140 Hz, width of a scanned beam @ 2 cm. The laser used for this treatment was manufactured by Clean Optical Laser System, CL 120 Q-1064 nm Nd:YAG. These are the specifications of this system:

Figure 3. Scraped coating and corrosion before ablation, normal light at 25X.

2

Establishing parameters of the laser beam

EXPERIMENTAL

1. In situ sampling and recording: Samples were collected before and after ablation with scalpels and by vacuuming using removable HEPA filters. The sample sites were recorded by digital photography (using a Nikon D-300). 21 samples were collected. Images were taken in the first few minutes following ablation. The ambient conditions at the time of sampling were 28°C, RH 80%. The samples were mounted as received in the lab on glass slides for uv and visible light examination under the microscope (Zeiss Stemi 11 and Leica Laborlux using uv-fluorescence). A redundant set of particles from samples 1–10 were placed on Al stubs with carbon adhesive, coated with carbon, and then analyzed in a JEOL 6740LV

− − − − − − −

Scan width (stylus): 1–25 mm Scan frequency: 50–150 Hz, Variable laser pulse frequency: 8–35 kHz. Pulse energy: 3–15 mJ Pulse duration: 5 nanoseconds Spot size: 0.5 mm Focal diameter: 400 μm.

This system uses a fiber optic cable to transfer light energy from the laser unit to the hand piece. The length of the fiber is of extreme importance because it allows a conservator to work freely without moving the laser unit frequently.

48

LACONA8.indb 48

11/10/2010 5:36:18 PM

The system had a 30 m fiber optic cable that allowed conservators to work during the entire project without need for moving the unit from a truck. A portable generator was used to supply power to laser. 2.2

Advantages of the use of lasers

Traditional chemical treatments require use of various chemical solutions. The most effective in removing Incralac are combinations of the following solvents: toluene, acetone, xylene, with some ethanol and methylene chloride. All of the above solvents present health hazards for the conservators and contaminants for the environment. According to EPA guidelines, all chemical runoff should be collected and disposed of properly. The cost of collection and disposal of chemical runoff is usually as expensive as the treatment itself [2]. During laser treatment, all of the material that is ablated from the surface can be collected by a vacuum cleaner; there is no runoff or contaminated blasting material to collect. Laser cleaning provides a means of Incralac removal with low environmental impact. When comparing the time and material costs of chemical treatments with laser cleaning followed by patina and coating, the laser treatment may not be more expensive or time consuming. Previous work in Philadelphia and Chicago 2004–6 tests on different coatings, including “insoluble” or cross-linking high performance resins— urethanes, epoxies and fluorocarbons, verified the feasibility of clear coatings removal by laser ablation, rather than melting or burning [3–6]. For efficient cleaning rates with low environmental impact, ablation relies on fracturing, ejecting, ballooning, curling the clear coating. This degree of cleaning is impossible to achieve using chemical stripping and can be only duplicated by aggressive blasting of the surface.

3

Figure 4.

The laser cleaning process in progress.

Figure 5. Sample of ablated Incralac ‘ballooned’ on the bronze surface (not magnified).

microscope (brown/black) by iron staining, caused by deterioration of the ferric nitrate based patina applied in 2003. The samples imaged in Figures 6A and B and Figure 8 were collected on paper held under the ablated area during the ablation process. The coating is often fractured or ejected from the surface as particles smaller than 1 μm. Particles well below 800 nm, however, were not detected, contrary to the findings for vacuum-collected samples (see Fig. 8 below). The Figure 6A illustrates the size and shape of ablated particles, using the secondary electron detector in the SEM. Of particular note is the wide range of composition, which the individual particles typically show, with some particles indicating that multiple phases are present. Figure 6B, a backscattered electron image of the same particles, shows that the particles appear to be of differing brightness. In this SEM mode, the image is based on differences in the elemental composition of the particles, higher Z components appearing brighter than lower Z components. The large central particle, for instance, is comprised mainly of the coating resin and therefore it is of

DISCUSSION

A very wide range of particle sizes, shapes and compositions were observed in the samples collected in situ from the plume, both by gravity and by vacuuming. The wide variety of the particle size illustrates uneven response to the laser beam, many samples being in the millimeter range while others were of sub-micrometer size, as Figures 4–11 show. Figure 5 illustrates a common result of the ablation process in situ, where most of the coating is ablated while some remains on the surface and curls from thermal effects of the ablation process. Many such particles are millimeter size or larger. Lacquer scrapings before as well as after ablation were often found to be discolored under the optical

49

LACONA8.indb 49

11/10/2010 5:36:18 PM

Figure 6A.

SEI image.

Figure 6B.

BEC image.

Figure 7. SEI image illustrating that large particles of the organic coating are often removed/ejected from the surface with little change in aspect.

Figure 8A. BEC image showing the range of particles collected on paper placed under the area being ablated. The FOV is 100 × 150 micrometers.

low brightness in the BEC image. Conversely, the particle in the upper right corner indicates that it is comprised of higher Z elements than the central particle, with some very bright small particles of very high Z composition superimposed. Figure 7 presents another common feature of the coating debris collected after ablation. There is little evidence of thermal absorption from the beam—no curling, balling up, melting—and no porosities caused by gas evolution. In other words, the coating resin often appears to manifest mechanical fracture as one of the causes of its disbonding from the substrate. Clearly the state of aggregation of the substrate and its interaction with the laser beam plays a role in dislodging the overlying lacquer film. Figure 8A displays sub-micrometer to 20+ micrometer particles collected by gravity under the surface on which the beam was focused. Figure 8B gives a typical EDS result for such particles. Vacuum collection of the ablated coating revealed a number of very small spherical particles

Figure 8B.

Typical spectrum of a particle shown in 8A.

that had not been observed in gravity—collected samples. Figure 9A below shows small spherical particles ranging from 400 nm–1 μm. The chemical composition of those small particles is illustrated in Figure 9B, i.e., an EDS spectrum (PMAS 3523) which shows carbon to be the main element constituent in the ∼400 nm particle on which the beam was focused.

50

LACONA8.indb 50

11/10/2010 5:36:19 PM

Figure 10. Illustrating the melting and gas porosity often found associated the inorganic phases underlying the clear coating.

Figure 9A.

Figure 9B. An EDS spectrum produced by focusing the bean on a ∼400 nm diameter particle in Fig. 8A.

Figure 11. Showing a fine, highly magnified zone of craters in material that is largely silicaceous.

The fluence of Q-switched Nd:YAG lasers is clearly sufficient to obtain an efficient cleaning rate with a low environment impact for a clear, acryloid based lacquer on a large scale outdoor bronze monument. As Figures 10 and 11 below show, the beam interaction can result in many of the features commonly found with corrosion removal by laser—localized melting, spalling and cratering and gas-induced porosity or tunneling [3,5, 8–10]. In this particular case, lacquer failure and corrosion of an underlying layer of weathering/ repatination resulted in the fortuitous removal of much unwanted chloride corrosion with the lacquer without measurable alteration of the underlying metal. The mechanisms by which the lacquer was removed appear to have varied considerably, likely a function of the varying focal distance of the beam over the uneven topography of the corroded bronze surface. Ultimately the laser cleaned to a relatively stable brochantite and cuprite rich surface, with little residual chloride.

Critical to this specific project and to lacquer removal from outdoor monuments in general is the differential absorption of the inorganic layers underlying the organic coating, since for many clear lacquer resins, substantial, direct absorption of the beam energy is not expected. In particular, the absorption of the laser beam by hydrated corrosion products (especially basic copper sulfate and chloride species in the corrosion crust) results in some discoloration of the corrosion/patina, usually with the formation of cuprite as the corrosion is decomposed—at least partly—into gaseous byproducts. The persistence of the cuprite thus created appears to depend on the fluence and the total dwell time of the beam, as well as the air flow affecting the plume. Preliminary experiments aimed at characterizing the reactions responsible for the lacquer disbonding according to corrosion species showed quite different rates of lacquer removal and of discloration:

51

LACONA8.indb 51

11/10/2010 5:36:19 PM

Cu, Cu2O << anhydrous CuCl, CuCl2 < brochantite, copper acetate monohydrate. The chlorides produced much whiter reaction products than anticipated in these first experiments. 4

Brygo, F. Dutouquet, C. Le Guern, F. Oltra, R. Semerok, A. and Weulersse, J.M. “Laser fluence, repetition rate and pulse duration effects on paint ablation,” Applied Surface Science 252 (2006), 2131–38. Burmeister, T. Meier, M. Haferkamp, H. Barcikowski, S. Bunte, J. and Ostendorf, A. “Femtosecond Laser Cleaning of Metallic Cultural Heritage and Antique Artworks,” LACONA V Proceedings, 2003, pp. 63–67. Dajnowski, A. Jenkins, A. and Lins, A. “The Use of lasers in Cleaning Large Architectural Structures,”APT Bulletin, Vol. XL (2009), no. 1, pp. 13–24. Lins, A. and Clare, T.L. “Evaluation of Fluorinated Protective Coatings for Outdoor Metals,” Metal 07, Proceedings of the International Conference on Metals Conservation, Amsterdam. 2007. Lins, A. PMA Report on the City Hall Tower Project to the Pew Charitable Trusts, Dec 2006; and A. Lins, PMA Report on the City Hall Tower Project to the City of Philadelphia, Jan. 2007. These reports included tests showing that the coating applied during the project could be removed in the future, after failure, by laser ablation. Madden, O. Abraham, M. Scheerer, S. and Werden, L. “The effects of Laser radiation on Adhesoives, Consolidants and Varnishes.” LACONA V Proceedings, 2003, pp. 249–54. Mittner, P. Weidemann, G. Haber, G. Conrad, W. and Gervais, A. “Laser Cleaning of Metal Surface— Laboratory Investigations,” LACONA V Proceedings, 2003, p. 83. MVA (Mary Miller) sample test results taken during CSOS treatment of Chase all Evil and Halt All Evil sculptures owned by Ripley Entertainment Inc., Florida, USA. Rode, A. Freeman, D. Baldwin, K.G.H. Wain, A. Uteza, O. and Delaporte, P. “Scanning the laser beam for ultrafast pulse laser cleaning of paint,” Applied Physics A (2008) 93;135–39. Statement based on the cost of chemical treatment of Equestrian Indian sculptures performed under supervision of A. Dajnowski in 2005.

CONCLUSIONS

Using a 120W multi-pulse Q-switched Nd:YAG laser, it has been demonstrated that efficient and low environment impact cleaning of a clear, acryloid based lacquer (Incralac) from a large scale outdoor bronze monument is feasible. In this particular treatment, lacquer failure and corrosion of an underlying, repatinated layer resulted in the fortuitous removal of unwanted corrosion with the lacquer. The coating was removed with some of previously applied patina, but a stable and well-adhered surface corrosion layer was left in place. This project demonstrated that it is possible to remove a deteriorated clear coating and if needed a patina applied during previous treatments. The work described above has shown that the analysis of ablation residues or by-products is affected by the collection process and is naturally more complicated in the field than in a laboratory. A very wide range of particle shapes and compositions were found, suggesting substantial inhomogeneity in substrate absorption of the beam and reflecting the difficulty of maintaining a strictly continuous level of fluence at the surfaces being cleaned under field conditions. It is clear that substrate absorption plays the critical role in determining the success of the ablation of a clear lacquer for the laser utilized in this project. REFERENCES Brochantite was created in a patination bath in the PMA laboratories and verified by XRD analysis of the surface conversion products, which showed a thin layer of cuprite underlying brochantite.

52

LACONA8.indb 52

11/10/2010 5:36:20 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

PROCON TT 49: Laser cleaning of ancient Egyptian wall paintings and painted stone surfaces B. Graue GZG, University of Goettingen, Germany

S. Brinkmann & C. Verbeek Neferhotep e.V., Cologne, Germany

ABSTRACT: The tomb of Neferhotep (TT49) is located in the Theban necropolis on the west bank of the Nile at the city of Luxor. The tomb is extensively damaged over time due to human inhabitation, the keeping of livestock, and the burning of mummies inside the tomb. Large areas of the precious historic depictions are covered with thick crusts of soot and layers of dirt, thus they are unreadable in many areas. As part of the conservation work at the Tomb of Neferhotep non-destructive cleaning methods are developed for the very fragile wall paintings and colored stone surfaces. Besides mechanical and chemical methods, laser cleaning is also tested. For conservation work in the tomb in Egypt a mobile device is required. Besides a flashlight pumped Nd:YAG-laser with Q-switch, a battery powered backpack fiber laser was tested. The application of a small size, light weight fiber laser is a new approach. First laboratory tests with the fiber laser showed regular and homogenous cleaning results including a high sensitivity towards pigments. Testing inside the tomb led to the definition of adequate parameters for a selective work process. Pre-testing, accurate examination of surfaces and analyses materials as well as careful conception and the skilled performance by well trained and experienced conservators are essential to any laser application. The implementation of laser cleaning with a fiber laser in combination with mechanical and chemical cleaning methods brought previously unreadable depictions of ancient Egyptian Art in Thebes to light. 1

THE PROJECT TT 49

The tomb is cut into the bedrock of the Gurna hills. A courtyard slopes down to the decorated façade and elaborately decorated ritualistic rooms: the vestibule, pillared room, and Figure niche. Underneath these rooms is the burial chamber where the sarcophaguses were placed—a roughly cut and undecorated passage sloping downwards. Walls and

The international conservation and research project on the ancient Egyptian tomb of Neferhotep (TT 49) has been in existence since 1999. Archaeological and egyptological research is carried out by Argentinean and Brazilian specialists. Conservation is being undertaken through the expertise of German conservators in an international co-operation with other scientists of different disciplines. The main objectives of the project are the stabilization and consolidation of the historic material, as well as the surface cleaning of wall paintings and painted stone surfaces. Condition and damage are surveyed, recorded and mapped; historic materials and techniques are analyzed; conservation methods and materials are developed. A site management plan secures climate monitoring, visitor regulation and the long-term preservation. (Graue et al., 2004). 1.1

The rock tomb of Neferhotep (TT 49)

The tomb of Neferhotep is located at the Old Egyptian Necropolis in Thebes, built in the reign of Pharaoh Eye around 1320 BC. (Davies 1933).

Figure 1.

Map of the location of the tomb.

53

LACONA8.indb 53

11/10/2010 5:36:20 PM

Archaeological findings and earlier references show evidence of a fire inside the tomb in the first half of the 19th century. At this time a number of the mummies, which had been placed inside the tomb, were burned. Due to this fire the decorated surfaces in the tomb are covered with thick crusts of soot. Depictions and hieroglyphics are unrecognizable; the properties of the historic material have been significantly altered. As well as this soot crusts built up surface tension leading to flaking off paint layer. (Brinkmann et al. 2006, 2007). Focusing on the research on non-destructive cleaning techniques for this work of art of great cultural importance, materials and methods are developed. These include conventional methods such as mechanical and chemical cleaning; further more laser cleaning has been successfully implemented on painted surfaces in the tomb.

Figure 2.

2

View inside the tomb.

SOILING AND CLEANING

2.1 Characterization of the soot ceilings of the ritualistic rooms are richly decorated with coloured figures, reliefs and wall paintings. These paintings on plastered walls and stone surfaces are thought to have had an organic binder (presumably gummi arabicum). Analyses show jarosite, yellow and red ochre, huntite, coal black, Egyptian blue and green. (Jägers 2009, Kutzke 2007) In some polychrome areas of the wall paintings a thick layer of a mastic-varnish was applied. 1.2

To develop specific cleaning methods, the composition of historic materials and overlying soiling is first analyzed, to find out about their chemical and physical properties as well as alteration and sensitivities. The very dense soot crusts have a specifically fatty and oily character due to the composition of the mummification materials and their high content of fats, oils and waxes. These large brownish-black soot conglomerates show a high content of ashes, parts of inorganic material (a.o. gypsum, calcite) as well as incompletely burned organic material. There is also evidence of an high content of carbonyls—as they are typical for resins and aged oils. (Läßig 2003, Herm, pers. comm.). The covering soot crusts not only mean an aesthetic constraint, they also change the physicalmechanical properties of materials, which occasionally leads to ongoing damage. Further more these highly porous soot layers posses a high level of adsorption for pollutants.

Deterioration and conservation

The interaction between several deterioration mechanisms and the pronounced cataclastic overprint, which disintegrates the stone into small pieces, have led to extensive damage of the historic substance. Deterioration phenomena range from black soot crusts on the decorated surfaces, insufficient adhesion of plaster, the flaking of paint layers, through cracking, crumbling and scaling of the stone, “splintering disintegration” of the bedrock, to the complete loss of the historic substance. (Graue et al., 2008). After a comprehensive survey a conservation plan was established, including the development of materials and techniques as well as a long-term preservation plan. Prior treatments focused on the stabilization and consolidation of the tomb’s structure, stone, plaster and paint layers. These include the fixing and injections of plaster and stone fragments, consolidation of fragile stone, plaster and flaking paint layers. Losses are filled with different lime mortars and the edges of fragile wall painting are stabilized. Besides the structural conservation, the cleaning of the very fragile surfaces presents another target.

2.2 Surface cleaning methods The surface cleaning aims to reduce the overlaying crusts of soot and dirt to put a halt to the ongoing damage processes and to gain a readability and understanding of the depictions in their historic context. Specific preserved patches remain to display the burning of the mummies as a part of the history of the tomb. Due to the very complex impairments different cleaning methods are tested: • mechanical cleaning (with brush, scalpel, erasers, ultrasonic, particle blasting, laser).

54

LACONA8.indb 54

11/10/2010 5:36:20 PM

• physically and chemically reacting solvents (water, acids, bases, soaps, chelating agent, enzymes, ionexchange, bleaching materials, organic solvents in combination with bases).

sity are again regulated through several parameters of the laser beam. A comparison of the two systems shows differences in pulse energy, pulse duration and repetitions rate but a similar power density.

Particularly in very sensitive areas of very little surface stability and high water solubility, as there is in the white background areas, where conventional methods are not applicable, it is expected that soot crusts can be gently reduced with a laser.

3.1.2 Laboratory tests Testing areas on the samples are defined through small masks (3 × 5 mm). The samples are treated with both laser systems starting with lowest energy level. In separate sample fields energy rises until a negative interaction such as color alteration or loss of material occurs. After testing the sample areas are examined with a microscope looking for alterations (color, material loss, surface alteration such as melting etc.). The testing is firstly to detect if in general a selective reduction of the soot crust without damage to the supporting surface (plaster, limestone, paint layer) is possible and secondly to define parameters. Power density on the Nd:YAG-laser can be regulated by three parameters: flashlight current (E), repetitions rate (F), optic (O). On the fiber laser power density is defined through five parameters: average laser Power (P), Scanning Width (SW), Scanning Frequency (SF), Repetitions rate (F), Number of Passes (NP).

3

LASER CLEANING

3.1

Pre-testing

Pre-testing in a laboratory was carried out with two different laser systems on samples of soiled lime stone, a fragment of plaster with paint layer, colored clay shards and a soot covered sample plate with a reconstruction of an ancient Egyptian painting. 3.1.1 Tested laser systems Applied for testing were two laser systems. Besides a flashlight pumped Nd:YAG-laser with Q—Switch, that is commonly used in conservation, a fiber laser was included in the tests. In comparison with a Q-switched Nd:YAG-laser with pulse durations of <10 ns, the fiber laser displays a high repetition system with longer pulse duration (100 ns). This battery powered fiber laser weighs 12 kg, hence it is much smaller than the Nd:YAG-laser consisting of three heavy units (Laser source, power supply, cooling unit). The ablation process is determined through the wave length of the emitted laser beam, the intensity, or power density and time of interaction. (Wiedemann & Kusch 2002) Intensity or power den-

3.1.3 Results A selective reduction of the soot crust is possible with both laser systems. On undecorated limestone samples and plaster fragments covered with a thick soot crust, it can already be reduced with a very low power density of 12 MWcm−2. However a yellowish surface remains. Applying chemical cleaning after the non-destructive laser treatment, the cleaning result can be improved. Further laser treatment to remove the yellow layer would result in damage to the surface of the object. Testing on a red colored soot covered plaster fragment shows the differences when applying the two systems. With the flashlight pumped laser no suitable parameters can be found for a nondestructive reduction of the soot layer. With the fiber laser a reduction of the soot without damage to the paint layer is possible.

Table 1. Comparison of parameters of the flashlight pumped and fiber laser (Panzner unpub. 2006).

Laser type Parameter Pulse energy [mJ] Pulse interaction area [mm2] Working distance [mm] Pulse duration [ns] Repetition rate [Hz] Power density [W/cm2] Wave length [nm] Beam control

Flashlight pumped laser system (Nd:YAG-laser Fiber laser system NL 102) (CL20 Q) ≈100 >1

≈0.25 <0.01

>350 +/− 30

160 mm +/− 1 mm

<10 <20 ≈107

>100 40,000–115,000 ≈107, <2.5 × 107

1064 With hand

1050–1060 Scanning line, with hand

Figure 3.

cleaning test on a fragment of plaster.

55

LACONA8.indb 55

11/10/2010 5:36:20 PM

Figure 4. Moving the fiber laser parallel to the surface— with a provisional nozzle.

Figure 5. Sample fields in the tomb—sample fields e and f partly discover a red decoration line.

The blue pigment on the clay shard is very stable even at high power densities. With a power density over p’ = 40 MWcm−2, it turns gray and de-colorizes. Both systems have a very diverse application for the user: with the Nd:YAG-laser single shots are placed, while the scanning fiber laser is moved parallel to the object surface. The cleaning result is very homogenous with little expenditure of time. 3.2

3.2.2 Results Working with the fiber laser a reduction of the soot layer is possible on the undecorated as well as decorated stone surfaces and wall paintings in the tomb of Neferhotep. On the stone surface the soot crusts can be thinned or reduced leaving a yellowish transparent surface layer. In a second step this layer can be chemically reduced, followed by a further laser treatment. In many areas a combination of chemical and laser cleaning brings about the best results. On undecorated stone surfaces a poultice with chemical solvents is first applied to reduce the organic soluble components of the soot crust. Afterwards this area is worked with the laser. In areas of decorated stone surface the fast evaporating solvents are applied with a thin paper tissue for a very short time (app. 2 minutes), followed by a laser treatment. The repeated alternation of these methods gives the opportunity of precisely controlling the process and ends in very good results. Extremely thick and dense soot crusts cannot be removed from the stone surface without material damage, since the oily and fatty components of the soot have penetrated into the porous system of the rock. Particularly good results show on the white background paint layer, where a self-regulating work process and the selective ablation of the soot layer without interference with the underground are possible. While conventional methods damage the sensitive surface, there are no losses after laser treatment. Similar to the treated stone surfaces a microscopically visual thin yellow-transparent layer can be observed. (Sobott et al., 2002, Wiedemann & Kusch 2002, Herm 2009, Jägers 2006, 2009). The cleaning tests in polychrome areas show varying results. Reducing the soot from red painted surfaces with the laser does not lead to any loss or damage of material (s. Fig. 5). However too many passes with the laser (NP) over the sample field could potentially cause a reduction of the red pigment layer.

Testing on the object

Due to the very good results of the fiber laser in the pre-testing, the small and mobile backpack device is employed in Egypt for the tests in the tomb. 3.2.1 Laser cleaning tests in the tomb Tests start on undecorated soot covered areas (limestone and plaster) to ascertain parameters for the reduction of the dirt layer. The surface is treated through masks of 5 × 10 mm. Each sample field is examined with microscope and documented before and after treatment. Micro-samples of the treated surface are evaluated through a variety of analytical methods (light microscopy, SEM, FT-IR, XRF and micro chemical analyses). After determining parameters for the nondestructive cleaning of the undecorated surfaces, polychrome areas are also tested. To define the power density, the following parameters are documented: Table 2. Documented parameters (CL20 Q) of the testing in the tomb. Abbreviation

Parameter

P F

Average laser power (2.5–20 Watt) Repetitions rate (40–115 kHz, scale 10–0) Scanning frequency (50–300 Hz, scale 0–10) Scanning width (0–63 mm, scale 0–10) Number of passes, movements with the laser beam over the sample area

SF SW NP

56

LACONA8.indb 56

11/10/2010 5:36:21 PM

Table 3. Parameters for the reduction of a dense soot crust with the fiber laser CL20Q.

Surface

P

F

SF

SW

W

kHz Hz

mm

40

300

30

5

40

300

30

4

40 40

300 300

30 30

18 4

40

300

30

2

Undecorated limestone 10 Decorated limestone 5 White paint layer 10 Polychromic 10 Red pigment 18th dynasty 5

NP

Power density MWcm−2

∼50 Figure 6.

Black paint layers underneath a black soot crust are a specific case for laser cleaning due to similar absorption spectra. Covering the black painted areas with Cyclododecan works successfully, since the temporary protection hinders the laser light for ablation for a short time. Limits of laser cleaning are also observed when the reduction of soot from painted areas with a blackened varnish is tested. A laser treatment of varnished areas is not possible. In general it can be said that a self regulated ablation process can not be achieved due to the very heterogeneous composition of painted surfaces and soot layers. Therefore it is of great importance to work very sensitively so as not to harm or damage the paint layers. The painted surfaces of decorated stone figures, reliefs and wall paintings with filigree structures of different colors mean a challenge to laser cleaning and the operator. Limiting parameters have to be found, that prevent any risk of damage to the materials. The skilled user has to react immediately to any change in the circumstances of the ablation conditions. The determined parameters for the backpack laser offer a reduction of the soot layers and crusts step by step, without damaging underlying colors. The very thin yellow transparent layer, which can be seen on stone and wall painting surfaces after laser cleaning, is analyzed. The exact characterization of composition, cause and effect should help to exclude any damaging or irreversible alteration of the historic surface due to laser treatment. Analyses of laser treated samples of the white background, which show a yellowish brownish layer, detect an organic or oily substance. This layer can also be observed on untreated reference samples of the white background with a black soot crust. These results suggest that the yellow coat could be residues of the oily fatty soot crust left on the surface. (Jägers 2006, 2009) A final evaluation of these analyses is due, since the analyses on

Before and after cleaning.

Figure 7. Working with the backpack laser inside the tomb.

similar samples of decorated stone surface have yet to be done. 4

CONCLUSION

In the context of the conservation and research project it was possible to compare two laser systems: the Nd:YAG-laser (NL 102) common to conservation with a new-generation mobile backpack fiber laser (CL20 Q). For the work in Egypt the backpack laser system is chosen due to the very good results obtained in the pre-testing and its high mobility and portability of a small size device. In the tomb laser tests are carried out on different surfaces. On undecorated and painted stone surfaces soot crusts can be reduced with the fiber laser. A combination of cleaning methods by alternating the application of chemical paper tissue poultices and laser cleaning was found to be favorable. Convincing results are revealed by the tests on the white background of the wall paintings. Through a self-regulating ablation process the soot layer can be gently and gradually reduced. Also polychrome areas of the decorated surfaces can be cleaned with the fiber laser. A cleaning process without damage to the surface is very difficult. Due to the heterogeneous character of

57

LACONA8.indb 57

11/10/2010 5:36:21 PM

Brinkmann, S., Graue, B. & Verbeek, C. 2007. Wiederlesbarmachung altägyptischer Darstellungen zu ihrer Interpretation. unpubl. report. research project 2004–2006. Köln: Neferhotep e.V. Davies, N.G. 1933: The Tomb of Nefer-Hotep at Thebes. Vol. I. New York: Metropolitan Museum of Art. Graue, B., Verbeek, C. & Brinkmann, S. 2004. PROCON TT 49: Conservation Project at the Tomb of Neferhotep in Thebes. In: Proceedings of Building Materials of Egyptian Monuments. International Forum I and II: 118–124. Cairo: SCA. Graue, B., Kordilla, J. & Siegesmund, S. 2008. Stone deterioration and conservation of the ancient Egyptian tomb of Neferhotep (TT 49) in Thebes (Egypt). In: Lukaszewicz & Niemcewicz (eds), 11th Interntional Congress on deterioration and conservation of stone: 1231–1238. Torun. Herm, C. 2006. Untersuchungen im Rahmen des Forschungsprojektes „Wiederlesbarmachung altägyptischer Darstellungen“. unpubl. report. Dresden: Labor für Archäometrie, HfBK. Herm, C. 2009. Untersuchungen im Rahmen des Forschungsprojektes “Wiederlesbarmachung altägyptischer Darstellungen”. unpubl. report. Dresden: Labor für Archäometrie, HfBK. Jägers, E. 2006. Untersuchungen im Rahmen des Forschungsprojektes “Wiederlesbarmachung altägyptischer Darstellungen”. unpubl. report. Bornheim. Jägers, E. 2009. Untersuchungen im Rahmen des Forschungsprojektes “Wiederlesbarmachung altägyptischer Darstellungen”. unpubl. report. Bornheim. Kutzke, H. 2007. unpubl. report. Cologne: Technical University. Läßig, A.-K. 2003, Konservierung und Restaurierung des “Epitaph für Johannes Wessels”, 1619, aus der Kirche St. Georg zu Wiek / Rügen. unpubl. Dresden: HfBK. Panzner, M. 2006. unpubl. report. Dresden: FraunhoferIWS. Sobott, R., Neumeister, K. & Seidel, H. 2002. Laserstrahlreinigen von Naturstein und naturwissenschaftliche Untersuchungen. In Wiedemann & Kusch (eds), Laserstrahlreinigen von Naturtstein. Stuttgart: Fraunhofer IRB. Wiedemann, G. & Kusch, H.-G. 2002: Der Laserstrahl als Werkzeug für den Restaurator. In Wiedemann & Kusch (eds), Laserstrahlreinigen von Naturstein: 25–4. Stuttgart: Fraunhofer IRB.

the materials a self-regulated ablation process is not possible. Often the laser treatment only works successfully in combination with other conservation and cleaning techniques. In some areas a mechanical or chemical preliminary treatment is necessary. The ablation process itself has to be regulated by the conservator. This requires a precise evaluation of each situation and an exact knowledge of the object as well as comprehensive experience in operating the laser device. The pre-testing serves for the definition of parameters for a damage free cleaning. These parameters are always specific for the object as well as the laser device; hence they are not directly transferrable to other objects or other laser devices. It is of crucial importance that analyses and pretesting are carried out in advance to every laser treatment in conservation. A correct and careful application of laser cleaning opens new possibilities for work on very fragile and soiled historic surfaces and to preserve them. As well as the ongoing structural conservation and long-term preservation, the development of cleaning techniques for the tomb of Neferhotep proceeds successfully. During the campaigns in Egypt from 2004 to 2009 broad areas of soot covered surfaces in the tomb have been cleaned by the application of a combination of conventional mechanical and chemical cleaning methods with laser cleaning. Depictions and decorative details previously obscured by soot can be seen once more. ACKNOWLEDGEMENT We would like to thank Gerda-Henkel-Foundation for the financial support, through which testing of laser cleaning was made possible for this project. Clean—Lasersysteme, Herzogenrath and Fraunhofer Institut für Werkstoff- und Strahltechnik (IWS), Dresden helped with their technical and institutional support. Our gratitude goes to the head of the project, Prof. M. Violeta Pereyra, University of Buenos Aires and the members of Neferhotep e.V.

MATERIALS Fiber laser: Laser CL 20 Q Backpack. Fa. Clean-Lasersysteme GmbH, Herzogenrath, Germany. Flashlight pumped Nd:YAG–laser with Q—Switch (Type NL 102) Fa. BM- Industries, Evry, France.

REFERENCES Brinkmann, S., Graue, B. & Verbeek, C. 2006. Die Erforschung und Konservierung der altägyptischen Grabkammer des Neferhotep in Theben. In: In & out. Festschrift zum 20-jährigen Jubiläum: 29–34. Köln: Institut für Konservierungswissenschaften.

58

LACONA8.indb 58

11/10/2010 5:36:22 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

The influence of paper type and state of degradation on laser cleaning of artificially soiled paper S. Pentzien, A. Conradi & J. Krüger BAM Federal Institute for Materials Research and Testing, Division VI.4 Surface Technologies, Berlin, Germany

ABSTRACT: Lasers can be a supplemental tool for restorers to overcome some of the limitations of traditional dry cleaning techniques for works of art on paper. The laser working range has to be optimized allowing for safe removal of contamination and limitation of damage to the substrate. This paper addresses the influence of paper type and state of degradation on laser working range. Three types of new paper (pure cellulose, bleached pulp paper, rag paper) were degraded and characterized with respect to their degree of polymerization. Laser-induced damage thresholds of new and degraded paper were determined using SEM and viscometry. Additionally, artificially soiled model samples were made using two kinds of soiling, namely pulverized charcoal and soot-blackened standard test dust. Cleaning thresholds of soiled paper samples were evaluated. A working range for all combinations of paper and soiling between 0.05 J/cm2 and 0.5 J/cm2 was found for the application of 8-ns laser pulses at 532 nm wavelength. 1

INTRODUCTION

photon energies is the potential to induce a direct photolysis or photo-oxidative degradation of cellulose (Kolar et al. 2000a). For the 308 nm laser treatment, photo-oxidative degradation of the paper substrate was accompanied by an increase in the content of oxidized groups and a severe decrease in the Degree of Polymerization (DP). In the recent past, Nd:YAG lasers with a fundamental wavelength at 1064 nm and a second harmonic at 532 nm were utilized for cleaning purposes avoiding the problem of direct bond breaking of the organic substrate material. However, 1064 nm irradiation of Whatman filter paper can lead to the formation of inter- and intra-molecular cross-links of ether origin resulting in an increase of the DP (Kolar et al. 2000a). Additionally, color changes (yellowing) of the cellulose substrate as a result of thermal degradation of soiling materials were found after laser cleaning of artificially soiled paper samples with laser energy densities of 1 J/cm2 and lower (Strlič et al. 2003). Laser treatment of Whatman filter paper and bleached sulfate pulp with the second harmonic radiation of the Nd:YAG laser (532 nm) and energy densities below 0.86 J/cm2 caused no degradation of paper (Kolar et al. 2000b). Several investigations confirmed the finding that 532 nm is the preferable wavelength for cleaning of paper (Kaminska et al. 2004, Rudolph et al. 2004, Kaminska et al. 2007). Recently, results of 8-ns laser cleaning of artificially soiled Whatman filter samples were

Laser cleaning can be considered as a wellestablished technique in stone and metal conservation, while laser treatment of complex organic materials like paper, parchment, and textiles is still not fully developed for application in conservation workshops. Paper is a medium for conveying text and images. As soiling interferes with the perception of information paper cleaning may be needed. In principle, laser cleaning as a non-contact method offers advantages due to selectivity, possible in-situ monitoring of the cleaning process, avoidance of solvents, versatility and reliability of the laser technique and control of laser wavelength, pulse duration, pulse repetition rate and spatial beam profile. Laser cleaning can be a supplemental technique for restorers especially in cases when conventional cleaning is impossible, e.g. for fragile works of art with low mechanical stability. The laser working range has to be chosen in such a manner that the contamination can be removed without any deterioration of the paper matrix material and without inducing damage in the long-term. The laser working range has to be chosen in such a manner that the contamination can be removed without any deterioration of the paper matrix material. More than ten years ago, excimer lasers emitting wavelengths in the ultraviolet spectral region were used to clean paper (Friberg et al. 1997, Kautek et al. 1998). The major drawback of high single

59

LACONA8.indb 59

11/10/2010 5:36:22 PM

compared with conventional eraser cleaning. Satisfactory laser cleaning effects without visible mechanical damage and yellowing of the paper were reached by employing a pulse repetition rate of 500 Hz and Gaussian beam diameters of about 0.1 mm (Krüger et al. 2008). Even at 532 nm, discoloration of the substrate material after laser cleaning was reported. The discoloration increases with increasing density of surface soiling (Strlič et al. 2005). Furthermore, Strlič et al. demonstrated (for 1064 nm wavelength) that a lowering of the pulse repetition rate from 10 Hz to 1 Hz and a reduction of the laser fluence (from 0.1 J/cm2 to 0.05 J/cm2) caused a reduction of the discoloration phenomenon. The work was done with laser beam diameters of 8 mm (1064 nm) and 5 mm (532 nm), respectively. In the present work, the dependence of laserinduced damage thresholds on paper type (pure cellulose, bleached pulp paper, rag paper) and its degradation are evaluated. Artificially soiled model samples of all paper types using two kinds of soiling were produced. A systematic laser cleaning study reveals laser working ranges for the different model samples.

2

Viscometric determinations of Degree of Polymerization (DP) were performed as described by Strlič et al. (2005) at the University of Ljubljana, Slovenia. DP values before and after accelerated degradation of unsoiled papers samples are displayed in Table 2. Artificially soiled model samples were prepared utilizing two kinds of soiling. Pulverized charcoal and soot-blackened standard test dust SAE J 726 fine (Particle Technology) were used. For each paper type, each kind of soiling was brushed on the paper. The soiling dust was vacuum sucked into the paper for 5 min. Afterwards, brushing and vacuum treatment were simultaneously performed for two minutes to achieve an adherence between the artificial pollution and the paper. As an example, Figures 1 and 2 depict Whatman filter paper soiled with charcoal and soot-blackened standard test dust, respectively. The homogeneity of soiling of model specimens was controlled on each sample using a multispectral imaging system (MUSIS 2007, Model D-HFA-12, Art Innovation) with lateral resolution <0.5 mm. Differences of lightness measurements Table 2. Degree of Polymerization (DP) of paper before and after accelerated degradation. KL = Käßberger Long, KS = Käßberger Short, ISO = ISO 5630-3.

EXPERIMENTAL

2.1 Sample preparation and characterization Three types of new paper were selected: – cellulose (Whatman filter paper No. 1), – bleached pulp paper (No. 2, LANA, Aqua Classic), – rag paper (Arches, Artistico, Vang). Paper samples were artificially degraded according to the procedure described by Käßberger (1998) and the relevant standard (ISO 1996) in a climatic chamber WK11 180 (Weiss Umwelttechnik GmbH). Table 1 lists the essential parameters of the different methods of accelerated degradation. The dynamic degradation methods included moist heat treatment at 80°C as well as 200 cycles (Käßberger long) and 48 cycles (Käßberger short) of relative humidity between 30% and 90%, respectively. During static degradation, the samples were kept at 80°C and 65% relative humidity. Table 1. Method

Paper

DP As received

DP KL

DP KS

DP ISO

Whatman no. 1 No. 2 LANA Acqua classic Arches Artistico Vang

3550 1980 2380 Not soluble 1970 2370 3000

1860 1800 2030

2710 1900 2030

3140 1990 2290

1750 1530 2970

1840 1980 3050

1910 2070 3570

Procedures of artificial degradation of paper. Type

Parameter

Käßberger long Dynamic 80°C, 25 days, 30–90 RH% Käßberger short Dynamic 80°C, 6 days, 30–90 RH% ISO 5630-3 Static 80°C, 6 days, 65 RH%

Figure 1. Overall view of a Whatman filter paper No. 1 soiled with charcoal.

60

LACONA8.indb 60

11/10/2010 5:36:22 PM

Figure 2. Overall view of a Whatman filter paper No. 1 soiled with soot-blackened standard test dust.

Figure 3. Overall view of a bleached pulp paper (LANA) after laser treatment with various energy densities and numbers of pulses per spot.

in the visible spectral region (visible reflection mode of MUSIS 2007) of ±10% maximum were measured.

3

2.2

3.1

Laser treatment

RESULTS AND DISCUSSION New and aged paper: Laser-induced damage threshold

For pure cellulose (Whatman filter paper No. 1), bleached pulp paper (LANA, No. 2), and rag paper (Vang) multi pulse Laser-Induced Damage Thresholds (LIDT) of new and accelerated aged samples were determined using SEM. To examine the worst case conditions, specimens after artificial ageing according to Käßberger (long) were used. Figures 4–6 show Scanning Electron Microscope (SEM) pictures of Whatman filter paper No. 1 in its initial state (Figure 4), after laser treatment of a new sample at the damage threshold (Figure 5), and after laser treatment of an aged specimen slightly above LIDT (Figure 6). Figures 7–9 depict SEM pictures of the bleached pulp paper LANA in its natural appearance (Figure 7), after laser treatment of new LANA at the damage threshold (Figure 8), and after laser treatment of aged LANA marginally above LIDT (Figure 9). Whatman filter paper and bleached pulp paper were illuminated with high laser fluences and multi pulse conditions to determine a forbidden range of laser parameters. The appearance of new and artificially aged paper surfaces after such a massive intervention is displayed in Figures 5, 6 and 8, 9, respectively. Mechanical damage of the paper material is observed. The damaged surfaces can be clearly distinguished from intact paper substrates (Figures 4 and 7). Figure 10 summarizes multi pulse LIDT values for five paper substrate materials. Firstly, the damage thresholds of new paper range from about 11 J/cm2 for pure cellulose (Whatman) to 0.6 J/cm2 for rag paper (Artistico). i.e., LIDT values differ by more

The experiments were performed with a laser cleaning system described in detail by Kautek & Pentzien (Kautek & Pentzien 2005). A Q-switched Nd:YAG laser was operated at 532 nm wavelength at a pulse duration of 8 ns and a repetition rate of 500 Hz. Pulse energies <2 mJ were measured by means of an energy meter (Nova, Ophir). The spatial beam profile was Gaussian with a beam radius (1/e2) of 100 μm on the sample surface. The laser beam was scanned over the sample through a remote computer control system. With a camera, the laser action was monitored on a computer screen. By a variation of the laser scan rate, number of scans and repetition rate, the number of illuminating laser pulses N on each point of a selected area can be varied. In this work, N values between 5 and 100 were applied. Normally, 3 mm squares were scanned with a fixed set of laser parameters. Figure 3 shows a bleached pulp paper sample with different lasertreated areas for the determination of the laserinduced damage threshold of the pure substrate material. The whole laser-processing compartment fulfils laser class I conditions. An exhaust system is integrated to remove gaseous and solid laser ablation products. The samples were inspected by means of optical (Eclipse L200, Nikon) and scanning electron microscopy (Stereoscan 180, Cambridge, accelerating voltage 10 kV). If any change of the fiber structure of the samples was observed by a scientist, laser-induced damage threshold was reached.

61

LACONA8.indb 61

11/10/2010 5:36:22 PM

Figure 4. received.

SEM picture of Whatman filter paper, as

Figure 7. SEM picture of bleached pulp paper (LANA), as received.

Figure 5. SEM picture of new Whatman filter paper after pulse laser treatment with 10.1 J/cm2 (N = 100).

Figure 8. SEM picture of new bleached pulp paper (LANA) after pulse laser treatment with 1.9 J/cm2 (N = 100).

Figure 6. SEM picture of artificially aged Whatman filter paper after pulse laser treatment with 11.2 J/cm2 (N = 100).

Figure 9. SEM picture of artificially aged bleached pulp paper (LANA) after pulse laser treatment with 2.0 J/cm2 (N = 100).

62

LACONA8.indb 62

11/10/2010 5:36:23 PM

Figure 11. Degree of Polymerization (DP) for different papers. Grey: new paper. Black: aged paper (Käßberger long).

Figure 10. Laser-Induced Damage Thresholds (LIDT) for different papers. N = 10. Grey: new paper. Black: aged paper (Käßberger long).

than an order of magnitude. The large variation of LIDT values can be a result of a different absorption of laser light in varying paper matrices, fillers and gluing. The basic materials of Whatman filter paper, rag paper and bleached pulp paper are cotton cellulose, fabric fibers and wood fibers, respectively. Secondly, non-aged papers withstood higher laser energy densities than aged papers. For pure cellulose, LIDT of non-aged and aged samples were nearly equal while DP value of aged paper was only 53% of its initial value (Figure 11). For bleached pulp paper (LANA, No. 2) and rag paper (Vang), DP of 85% (LANA), 91% (No. 2) and 99% (Vang) of their initial values were measured after accelerated ageing and respective LIDT of about 80% of new paper were found. 3.2

Figure 12. Whatman filter paper soiled with charcoal. Laser-treated squares with different brightness as a result of varying laser parameters can be seen.

Artificially soiled model samples: Laser cleaning threshold

Table 3. Cleaning thresholds of pulverized charcoal on paper in dependence of the detection method. N = 100.

For a successful cleaning of soiled paper samples, a laser working range with respect to number of pulses per spot, energy density and (possibly) pulse repetition rate has to be specified. In subsection 3.1, damage thresholds of paper substrates were evaluated. Here, cleaning thresholds of the different soiling types are indicated. Figure 12 shows a model sample featuring a few laser-treated areas. Thresholds for the removal of test contaminations from paper substrates are listed in Tables 3 and 4. Cleaning threshold means the lowest change of the soiling detectable by the methods listed. The ablation of pulverized charcoal from paper substrates was detected with different techniques (Table 3). Depending on their sensitivity, cleaning thresholds vary by a factor of three. The highest sensitivity was reached with in-situ particle detection (Pentzien et al. 2008).

Cleaning threshold [J/cm2] Detection

Whatman p. Wood-pulp p. Rag paper

Naked eye 0.022 Microscope MUSIS 0.014 Colorimetric Particle emiss. 0.007 Dust monitor

0.022

0.022

0.014

0.014

0.007

0.007

Colorimetric and microscopic ex-situ analysis yielded cleaning thresholds of 0.01–0.02 J/cm2 for pulverized charcoal (Table 3) and 0.02–0.04 J/cm2 for soot-blackened standard test dust (Table 4), respectively.

63

LACONA8.indb 63

11/10/2010 5:36:23 PM

Table 4. Cleaning thresholds of soot-blackened standard test dust SAE J 726 fine (Particle Technology) on paper. N = 100. Cleaning threshold [J/cm2] Detection

Whatman p.

Wood-pulp p.

Rag paper

Naked eye Microscope

0.038

0.038

0.024

Figure 15. SEM picture of Whatman filter paper soiled with charcoal after pulse laser treatment with 1.75 J/cm2 (N = 5).

An increase of the energy density to 1.75 J/cm2 leads to a complete removal of the soiling. 4

CONCLUSIONS

For successful nanosecond laser cleaning of soiled paper at 532 nm wavelength, paper type and its state of degradation are of relevance. A systematic study revealed laser working ranges for different model samples representing essential characteristics of contaminated real world artworks. New paper samples consisting of pure cellulose, bleached pulp paper or rag paper were thermally aged and characterized by their degree of polymerization. Multi pulse laser-induced damage thresholds of new paper between 11 J/cm2 (pure cellulose) and 0.6 J/cm2 (rag paper, Artistico) were found. Aged papers withstood about 80% of the laser energy density compared to non-aged papers Cleaning thresholds for different types of contamination in a range of 0.01–0.02 J/cm2 (pulverized charcoal) to 0.02–0.04 J/cm2 (soot-blackened standard test dust) were obtained. A laser working range for all combinations (except degraded Artistico) of paper and soiling between 0.05 J/cm2 and 0.5 J/cm2 can be suggested.

Figure 13. SEM picture of Whatman filter paper soiled with charcoal.

Figure 14. SEM picture of Whatman filter paper soiled with charcoal after pulse laser treatment with 0.5 J/cm2 (N = 5).

ACKNOWLEDGEMENTS Figure 13 represents a scanning electron microscope picture of Whatman filter paper soiled with pulverized charcoal. The soiling particles and cellulose fibers are depicted. Figure 14 shows the surface of Whatman filter paper after laser cleaning with five pulses per spot and a moderate laser fluence of 0.5 J/cm2. The cleaning effect is obvious and only a few soiling particles can be identified.

The authors want to thank Danijela Pucko Mencigar, University of Ljubljana, Slovenia, for performing DP measurements and Birgid Strauß and Dr. Michael Bücker, BAM Federal Institute for Materials Research and Testing, Berlin, Germany, for taking SEM pictures and support of the degradation experiments, respectively.

64

LACONA8.indb 64

11/10/2010 5:36:24 PM

Kolar, J., Strlič, M., Müller-Hess, D., Gruber, A., Troschke, K., Pentzien, S. & Kautek, W. 2000b. Near-UV and visible pulsed laser interaction with paper. Journal of Cultural Heritage 1: S221–S224. Krüger, J., Pentzien, S. & Conradi A. 2008. Cleaning of artificially soiled paper with 532-nm nanosecond laser radiation. Applied Physics A 92: 179–183. Pentzien, S., Conradi, A., Krüger, J. & Wurster, R. 2008. Monitoring of the laser cleaning process of artificially soiled paper, In M. Castillejo, P. Moreno, M. Oujja, R. Radvan & J. Ruiz (eds), Proceedings “Lasers in the Conservation of Artworks VII” (LACONA VII): 345–351. London: Taylor & Francis Group. Rudolph, P., Ligterink, F.J., Pedersoli jr., J.L., van Bommel, M.R., Bos, J., Aziz, H.A., Havermans, J.B.G.A., Scholten, H., Schipper, D. & Kautek, W. 2004. Characterization of laser-treated paper. Applied Physics A 79:181–186. Strlič, M., Kolar, J., Ŝelih, V.S. & Marinček, M. 2003. Surface modification during Nd:YAG (1064 nm) pulsed laser cleaning of organic fibrous materials. Applied Surface Science 207: 236–245. Strlič, M., Ŝelih, V.S., Kolar, J., Kočar, D., Pihlar, B., Ostrowski, R., Marczak, J., Strzelec, M., Marinček, M., Vuorinen, T. & Johansson, L.S. 2005. Optimisation and on-line acoustic monitoring of laser cleaning of soiled paper. Applied Physics A 81: 943–951.

REFERENCES Friberg, T.R., Zafiropulos, V., Kalaitzaki, M., Kowalski, R.P., Petrakis, J. & Fotakis, C. 1997. Excimer laser cleaning of mold-contaminated paper: Sterilization and air quality considerations. Lasers in Medical Science 12: 55–59. ISO 5630–3:1996. Paper and board—Accelerated ageing— Part 3: Moist heat treatment at 80 degrees C and 65% relative humidity. Käßberger, M. 1998. Vorgänge im Papier bei dynamisch beschleunigter Alterung, PhD thesis, TU Graz. Kaminska, A., Sawczak, M., Cieplinski, M., Sliwinski, G. & Kosmowski, B. 2004. Colorimetric study of the post-processing effect due to pulsed laser cleaning of paper. Optica Applicata 34: 121–134. Kaminska, A., Sawczak, M., Komar, K. & Sliwinski, G. 2007. Application of the laser ablation for conservation of historical paper documents. Applied Surface Science 253: 7860–7864. Kautek, W., Pentzien, S., Rudolph, P., Krüger, J. & König, E. 1998. Laser interaction with coated collagen and cellulose fibre composites: fundamentals of laser cleaning of ancient parchment manuscripts and paper. Applied Surface Science 127–129: 746–754. Kautek, W. & Pentzien, S. 2005. Laser cleaning system for automated paper and parchment cleaning. Springer Proceedings in Physics 100: 403–410. Kolar, J., Strlič, M., Pentzien, S. & Kautek, W. 2000a. Near-UV, visible and IR pulsed laser light interaction with cellulose. Applied Physics A 71: 87–90.

65

LACONA8.indb 65

11/10/2010 5:36:24 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Laser cleaning studies for the removal of tarnishing from silver and gilt silver threads in silk textiles B. Taarnskov The National Museum of Denmark, Brede, Kgs. Lyngby, Denmark

P. Pouli IESL-FORTH, Bassilika Bouton, Heraklion, Crete, Greece

J. Bredal-Jørgensen The Royal Danish Academy of Fine Arts, School of Conservation, Copenhagen, Denmark

ABSTRACT: The conservation of silver and gilt silver threads with core made of silk represents a complicated problem, as it involves different materials with diverse physicochemical properties. The purpose of cleaning is to restore the tarnished silver and gilt silver surfaces to their previous brilliance while preserving the fragile silk fibers that form the base of the thread. The feasibility for cleaning with lasers is studied on model samples. Tests were performed using lasers in the ultraviolet (248 nm) and visible regions of the spectrum (532 nm) in order to define the optimal laser parameters for the removal of silver sulphide from silver. The use of ultra-short laser pulses was tested to minimize exposure of silk to heat and light. Laser radiation at 532 nm with pulse width at 150 picoseconds (ps) could potentially remove the tarnish layers, still side effects to both the silver and the silk components were observed under certain laser parameters and thus further investigation towards the development of an optimum laser cleaning methodology is undertaken. The best cleaning results on silver were achieved with 248 nm and 500 femtosecond ( fs) pulse duration, but the silk threads were severely damaged. Tests on historic gilt silver fringes showed good results, as the gloss of the un-corroded gold made it easier to obtain a clean appearance. 1

INTRODUCTION

2003, Turovets et al. 1998, Siatou et al. 2006). Longer wavelengths at 355 nm and 532 nm were able to clean the metal but resulted in side effects such as discolouring, caused by re-precipitation of corrosion products, and melting. Even longer wavelengths (at 1064 nm) had a lower cleaning capacity and caused too high temperatures leading eventually to severe discolouring (Degrigny et al. 2003, Lee et al. 2003, Siatou et al. 2006). To keep the temperature in the silver as low as possible the pulse durations should be short, i.e. a few nanoseconds, or even shorter in the range of several picoor femtoseconds (Siatou et al. 2006). In the case of silk, previous studies indicated that the minima possible alteration occurs upon irradiation in the visible spectrum (at 532 nm). In contrary, ultraviolet (UV) light will decrease the degree of polymerisation and thereby will cause loss of physical strength, while infrared (IR) light results into yellowing. The changes initiated by the laser treatment will catalyse continuous physiochemical changes in the silk fibres. (Belli et al. 2006, Reichert 1998, Lerber et al. 2005) and thus it is essential to avoid them to the most possible extent.

The conservation of silver and gilt silver threads that decorate brocade, embroidery and passementerie in museum collections is complex as it involves treatment of a variety of materials with diverse physicochemical properties. In this project, the cleaning intervention aims to restore the tarnished silver and gilt silver threads to their previous brilliance and luxury, while preserving the fragile silk fibers that form the base of the textile construction. Knowing the disadvantages involved with the use of traditional mechanical and chemical cleaning techniques, the feasibility of lasers was studied. The choice of using laser irradiation for cleaning tarnished silver offers several possibilities, given the exceptional control and spatial confinement that laser ablation of materials offers. Previous experience from studies aiming to investigate the applicability of laser radiation to clean silver and gilt silver threads in silk textiles showed that the best cleaning results were achieved using wavelengths at 193 nm or 248 nm (Lee et al.

67

LACONA8.indb 67

11/10/2010 5:36:24 PM

Pouli et al. 2008, Fotakis et al. 2006) have shown the unique advantages of cleaning with ultra-short pulses and this possibility was also examined in this study, in order to minimize any potential damage due to heat and light to the fragile silk substrate. This article describes the findings of the Diploma Thesis by B. Taarnskov from The Royal Danish Academy of Fine Arts, School of Conservation, Copenhagen 2009 (Taarnskov, B. 2009).

From the above it is obvious there are many obstacles and limitations in the laser irradiation of such fragile and complicated systems, which call for particular attention. A significant issue is that both the layer to be removed (silver sulphide) and the authentic surface (silver), have similar optical properties (i.e. absorption coefficients for each employed wavelength) and thus no “self-limiting” phenomena are operative. Additionally to this, the optical properties of the sensitive fragile silk core are very similar to the metal ones and given the ultra-thin nature of the metal thread and the geometry of the thread system (metal mounted around the silk) add further limitations to the laser cleaning procedure. To ensure a systematic approach to this multifaceted cleaning issue a series of laser cleaning studies on a variety of laser cleaning regimes were undertaken. Emphasis was given to use wavelengths that are highly absorbed by the material to be removed (silver sulphide) while their absorptivity by the core material should be as minimal possible. Visible (VIS) radiation at 532 nm (2nd harmonic of Nd:YAG lasers) is absorbed rather weakly by all the materials under investigation, while the UV radiation at 248 nm (KrF Excimer laser) has a rather high absorptivity by both the silver (as well as its sulphide) and the silk core. More specifically the estimated absorptivity of silk at 532 nm is minimal (about 2%) while in this wavelength silver absorbs weakly (in the range of 5%) and its tarnish product absorbs slightly higher (about 10%) (Bennett et al. 1970, Lerber et al. 2005, Panzner, M. et al. 2007, Wise & Coxe 1961). On the other hand, at 248 nm all materials absorb laser radiation significantly (80–85% for silver and 90–95% for silver sulphide). As for these two wavelengths no significant difference exists between the absorptivity of the overlayer (tarnish) to the original surface (silver) and the under-layer (silk core), thus a careful selection of the laser parameters is imperative. Towards this aim, it was decided to test these wavelengths on a series of reference and real samples in order to investigate whether it would be possible to remove efficiently the tarnish layers without affecting the ultra-thin structure of the silver thread and its sensitive silk core. To experience how these two wavelengths work in practice, a series of tests were performed on samples of new silver plates and silk threads as well as gilt silver fringe with silk core using lasers at UV and VIS wavelengths in order to define the cleaning threshold and the optimal laser parameters to remove silver sulphide from silver. A particularly important aspect to this exceptionally difficult cleaning intervention was the possibility to investigate also the influence of pulse duration to UV and VIS laser ablation. Extensive studies on industrial and medical, as well as, on Cultural Heritage (CH) applications (Burmester et al. 2005,

2 2.1

EXPERIMENTAL PART Test materials

Five different categories of test materials were included in the experiments giving supplementary information. Preliminary studies were carried out to define the effect of cleaning, extent of damage and the optimal laser parameters for cleaning. For this purpose reference un-corroded and artificially tarnished silver plates (0.1 mm thick) were used. In a second step, silk threads with no dye, bleach or weighting agents were tested in the similar range of parameters defined in the preliminary study. Reference and artificially tarnished silver strips (0.03 mm thick and 2 mm in width) were also tested to investigate whether the thickness of the silver foil has an influence on the impact of the laser. The strip has the same thickness, as the wrapping in the historic fringe, which is described in details below. The artificially tarnished silver plates have been treated in a chamber with sulphide vapour in a dry atmosphere. To create this climate powder of potassium sulphide (K2S) and a saturated solution of calcium chloride (CaCl2), were used which give a relative humidity at 45%. A fan produced circulation of the air. The materials were kept in the chamber for 1 month. The method is adjusted to this actual material using IEC-standard 68-2-46 (CEI/IEC International Standard 68-2-46). A combination of reference un-corroded and artificially tarnished silver strips mounted in cardboard frames together with the silk threads was used to investigate whether irradiation of the silver had a damaging impact on silk.

Figure 1. Samples mounted for the laser tests. From left: gilded silver threads with silk core from the fringe, combination of reference (un-corroded) and artificially tarnished silver strips together with the silk threads, and silk threads mounted alone.

68

LACONA8.indb 68

11/10/2010 5:36:24 PM

2.3

After these first tests, similar studies were carried out on fringes from the late 19th century, to define the threshold of cleaning and damage, and find the optimal laser cleaning parameters for this combined material. They are made of gilt silver strips around a yellow coloured silk core. The fringe is 0.8 mm thick made by two plied single threads of 0.3 mm in thickness. The gilded fine silver strips are 0.03 mm thick and 0.3 mm wide on a core, which is not twisted. The metal strips are corroded by silver sulphide tarnishing, which has penetrated the ultra thin gold layer.

The evaluation of the laser treated samples involved observation using an Optical Microscope (OM, Jenamed 2, transmitted, polarized light and Axiotech 100 HD) and a low vacuum scanning electron microscope (SEM, Jeol JSM-5310 LV), while tensile strength tests were performed on irradiated silk threads. In the optical microscope natural variations and changes caused by manufacturing as well as those induced by laser were detected. In the case of the gilt fringe, manufacturing degradation features were also observed. In the SEM the morphology of the silver and the silver gilt surface was studied, and quantitative analyses of the metals and corrosion products were carried out. Tensile strength tests were performed on the new silk threads only, because in the historic cores from the fringe the silk was degraded and too weak to be tested for tensile strength. The tensile testing machine was an Ingstron 5566 with load cell at 100 Newton. The tests were repeated for 10 samples, the temperature was 22°C and the RH was 56.5%. The samples were acclimatized in the climate for two weeks.

2.2 Laser irradiation parameters The laser irradiation schemes tested in this study span from UV to IR wavelengths and with pulse durations in the range of a few ns down to 500 fs. The specific information on the employed laser systems and irradiation parameters are shown in Table 1. Irradiation tests were performed on a spot-basis using 1, 5, 10, 30, 50, and in several cases 75 and 100 pulses. Irradiation on tarnish surfaces was stopped when a clean surface was obtained (determined by visible observation on the basis of efficient corrosion removal to the point that surface color and brightness was retrieved). Tests were also performed on reference un-corroded silver in order to establish the damage threshold of the authentic surface. Cleaning tests were performed in dry conditions as any supplementary wetting solution would risk damage to the silk and could initiate further or secondary corrosive schemes.

3

WaveOperative range length Pulse of fluences (nm) duration (mJ/cm2)

QS* Nd:YAG 532 (Palladio, Quanta systems spa) QS Nd:YAG 1064 (EKSPLA, SL312) QS Nd:YAG 532 (EKSPLA, SL312) KrF Excimer 248 (Lambda Physik, LPX200) KrF Excimer dye 248 system (Laser Lab. Göttingen)

8 ns

160–650 (tarnish silver) ≤920 (RU** silver)

150 ps

135–270

150 ps

30–300

30 ns

150–750 ≤1110 (RU silver) ≥400 (silk threads)

500 fs

40–105

RESULTS AND DISCUSSION

The Nd:YAG laser at 532 nm and 150 ps pulse duration was found to be able to remove corrosion from silver without visible damage to the silk. The short pulses at 150 ps caused the least changes to the silver surface morphology compared to the longer pulses at 8 ns. Discolouring of the silver is a potential problem, reported often in laser cleaning studies on silver objects (Degrigny et al. 2003, Lee et al. 2003), and there is a narrow interval between the threshold of cleaning and of damage. Discolouration may be produced when the total energy transferred by the laser is too low to remove the corrosion products but sufficient to initiate side effects (i.e. to oxidise the silver sulphide

Table 1. The laser systems and irradiation parameters employed for this study.

Laser systems

Evaluation of the irradiation tests

Figure 2. Reference un-corroded and tarnished silver plate treated with Nd:YAG 532 nm with 150 ps pulse duration at different fluences and number of pulses.

*QS = Q-switched, **RU = reference un-corroded.

69

LACONA8.indb 69

11/10/2010 5:36:25 PM

of damage, which were most clearly detected by loss in tensile strength of 60% and loss of extension of 50%. The results of the 10 single threads tested indicate that they have got different amount of exposure to the laser. The properties of the Excimer KrF laser at 248 nm and 30 ns pulse duration are different to the fs laser, and it is clearly not suitable for cleaning either silver or silk. Melting of the silver surface as well as the silk fibres is observed at energy levels lower than the threshold of cleaning. Cleaning attempts with the Nd:YAG at 1064 nm and 150 ps pulse duration was not satisfying. Discolouring was severe and melting occurred at fluences and pulse numbers lower than those required to clean. Table 2 summarises the above results showing the threshold values for each individual tested material at all tested cleaning regimes. It should be noted here that damage (melting) thresholds were determined on reference materials (silver plates/fringes and silk) and they indicate the maximum laser fluences that may be employed without damaging the substrate. On the other hand cleaning thresholds were determined on the tarnished samples and they specify the laser parameters for which removal of unwanted material is revealing a clean surface observed with the naked eye. For all lasers tested it was observed that changes in the morphology and discolouring occur at lower energy levels on the tarnished silver than on the clean silver. This is expected and can be explained by the theoretical values of the thermal and optical

to silver oxide). Depending on the thickness and the porosity of the silver oxide layer it appears yellow, purple or blue. By comparison, the Nd:YAG 532 nm with 8 ns pulse duration was not suitable to clean the silver plate, as the metal is discoloured and damaged at fluences and pulse numbers much lower than those required to clean. Common to both 532 nm irradiation regimes (at 8 ns and 150 ps pulse durations) is that they are gentle towards the silk threads, as no damages are seen. Furthermore, in the case of the gilded fringes (Figure 3), it is easier to assess when the fringe has been cleaned, as the un-corroded gold layer apparently acts as the desired level due to its compact surface and high reflectivity. Discolouring occurred only in a smaller scale. In this case for irradiation with 10 pulses at fluence values of 100 (±2) mJ/cm2 and 20 pulses at 80 (±2) mJ/cm2 a reasonably good cleaning is achieved without visual damage to the silk core or the gilded silver. Physical tests on the silk threads treated under these irradiation conditions showed no changes by the laser treatment. The maximum load before laser treatment was not reduced, and the single fibers broke on varying locations, but not in the laser treated area. The sulphur on the gilded silver surface of the fringe was quantitatively analysed in low vacuum SEM giving results of 6–9% before cleaning and 0.6% on the cleaned areas (using copper as calibration standard). The Excimer KrF laser at 248 nm and 500 fs pulse duration cleans the silver and the gilt silver in the fringe effectively and without discolouring or melting the silver surface. Irradiation in this regime with 30 pulses at fluence values of 100 (±2) mJ/cm2 gives a better visual result than any other test (i.e. in 532 nm). The sulphur on the silver surface was quantitatively analysed in SEM showing a reduction from 1% before cleaning to 0.06% after irradiation. Still, this laser (248 nm, 500 fs) causes severe damage to the silk fibres, as the threshold of damage seems to be far lower than the parameters needed for cleaning of the metal. Analyses of the silk threads after this treatment showed signs

Table 2.

Summary of the results. Cleaning threshold (mJ/cm2)

Laser systems

Materials

532 nm, 8 ns

Silver plate 650, 1 p* Silver fringe 360, 5 p Silk thread –

Damage threshold (mJ/cm2)

325–430, 1 p 360, 10 p No damage observed 532 nm, 150 ps Silver plate 300, 5 p 300, 30 p Silver fringe 150, 10 p 200, 20 p Silk thread – No damage observed 248 nm 30 ns Silver plate 680, 5 p 300, 1 p Silk thread – 200, 1 p 248 nm 500 fs Silver plate 60, 100 p and No damage 100, 30 p observed Silver fringe 100, 50 p Not defined Silk thread – 60, 50 p 90, 1 p 1064 nm 150 ps Silver plate >270, 50 p <180, 1 p Silver fringe >220, 50 p <135, 5 p

Figure 3. Gilded fringe un-treated (left) and irradiated with 20 pulses of F = 80 J/cm2 at 532 nm, 150 ps (right).

*p = pulse/s.

70

LACONA8.indb 70

11/10/2010 5:36:25 PM

Analyses of the silk threads were carried out before and after laser treatment. Analyzing the silk threads and the core from the fringe is complicated by natural variations in the silk fiber, which could be interpreted as laser induced damage. The historic core is degraded before treatment, for reasons we do not know in detail, and it is clearly shaped by pressure from the gilded silver wrapping. Variations and fractures are seen on the reference material as well as the treated samples treated with the Nd:YAG 532 nm lasers which means, that there is no laser induced damage on the new silk samples (Figures 4–5). On the other hand it was noticed that the historic core seem to have more fractures and split fibres in the laser treated samples than the reference core, indicating that the old silk is more vulnerable than the new. More analyses on a larger amount of test material are needed. The Excimer KrF laser at 248 nm and 500 fs pulse showed few incidents of a type of fracture and changes, such as uneven fiber surface, which is likely to be caused by the laser (Figure 6).

parameters of the two materials, silver (Ag) and silver sulphide (Ag2S), as presented by Turovets et al. Furthermore, it was noticed that the thin model silver strips are more sensitive towards the laser treatment than the thicker silver plate, as they show discolouring and melting at lower energy levels. Possibly this is due to the different manufacturing processes of the two samples leading to different structures in metal surface as grain size or degree of rolling and thereby stress. In any case, in all irradiation schemes followed in this study it is expected that heat diffusion is significantly restricted in the metallic surfaces tested and its numerical value lies far below the 30 μm, which is the thickness of the silver strips. Although the exact calculations of heat diffusion and flow on these systems are particularly precarious, as they are based on a series of idealistic assumptions, the heat diffusion length (L) is expected to be proportional to the laser pulse duration (τ) and may be calculated from the following equation: L = 2(k τ)1/2

(1)

where k is the thermal diffusivity of the material. In the case of silver (k = 169 10−6 m2/sec (Turovets, 1998) and for the longest pulse tested in this study (30 ns, for irradiation with the KrF laser at 248 nm) heat diffusion length is expected to be in the order of 4.5 μm which is significantly lower than the thickness of the irradiated samples. Of course for shorter pulse widths (8 ns for the 532 nm laser) this length is reduced, while in the case of ultra-short pulses of ps and fs duration, where additional phenomena take place, this length is expected to be even lower. Thus, the estimated thermal diffusion length in the metal upon irradiation at all tested lasers is minimal and highly confined; while no heat penetration to the underlying silk threads is anticipated. Indeed, tensile strength tests performed on the silk threads located underneath the silver strips indicated that these were affected to the same degree as the silk threads treated alone. From this we can conclude that the laser only affects the silk threads when they are irradiated directly by the beam. This issue must be taken into account when developing the laser cleaning methodology that would be adopted for this delicate cleaning intervention. It is suggested to modify the operative laser spot size accordingly to the size of both the gilded silver strip (0.3 mm) and the gap between the windings (often in the range of 0.1–0.3 mm), in order to avoid exposure of the silk to the laser. Similarly it may be necessary to adjust the laser scanning direction to the orientation of the fibers in the thread.

Figure 4. Observation in OM of a reference silk thread prior to irradiation.

Figure 5. Observation in OM of a reference silk thread upon irradiation at 532 nm of 150 ps pulse duration.

71

LACONA8.indb 71

11/10/2010 5:36:25 PM

core is a particularly delicate issue. The most vital concern is to choose the laser parameters in order to obtain a satisfying cleaning result, while keeping the applied laser energy below the limit of damage, as it was shown that the interval between cleaning and damage is indeed small. In this respect a reliable monitoring technique that would be able to indicate on-line the appropriate cleaning limits is crucial. Last but not least long-term issues must carefully addressed prior deciding on a laser cleaning strategy. Further studies in order to assess the behavior of the laser cleaned silver threads and their silk fibers with time are thus essential. Cleaning 3-dimentional objects of such a complex structure involves further considerations, as removal is limited only at the exposed parts of the surface. In this way, only limited parts of the clean metal are revealed. These areas are more porous and exposed to faster re-corrosion. Due to their particular sensitivity and susceptibility to corrosion, after cleaning the silver and gilt silver surfaces must be carefully protected. Placing the treated surfaces in a showcase with filtered air, keeping them together with an inhibiting agent or a sorbent, or seal them in air tight material, will potentially protect them against polluted air, so to avoid and prevent re-corrosion phenomena. Furthermore, through this study it was shown that upon any laser irradiation scheme alterations to the silk are inevitable. In the best case scenario (532 nm, 150 ps) these may be well restricted and minimized. Therefore we must always consider the reasons for choosing laser cleaning together with an evaluation of the advantages and the drawbacks, before we decide to proceed with the actual intervention. Finally, in order to develop a laser cleaning methodology to remove tarnish from silver and gilt silver threads in practice, it is necessary to take into account a series of aspects. Given the complexity of these objects their cleaning is a rather time-consuming, and thus expensive, process. Therefore to treat large objects in practice a methodology that will encompass the latest technological developments in order to achieve a safe, reliable and fast cleaning result is anticipated.

Figure 6. Observation in OM of reference silk threads upon 248 nm, 500 fs irradiation.

Evaluating the impact of the laser on the silk we must bear in mind that degraded fibres are more vulnerable to treatment than the new, and after being treated, they will degrade faster than fibres in good condition (Lerber et al. 2005). Every intervention will affect the materials to some extent, so it necessary to weigh the positive and negative aspects of laser cleaning. 4

CONCLUSIONS

The Nd:YAG laser at 532 nm and 150 ps pulse duration removed tarnish from silver gilt fringes satisfactorily, although tests with the same laser on the reference silver plates resulted in discolouration. According to the analysis the influence on the silk was minimal. The KrF Excimer at 248 nm and 500 fs pulse duration had the best cleaning result without any discolouration to the silver or the wrapping of the fringe, confirming that a shorter laser wavelength performs a more effective cleaning at lower fluences. However, an important drawback is the severe damage which was caused to the silk. Although this damage was not observed in the visual analysis, it was clearly detected in the tensile strength tests, showing a loss in strength of 60%, and loss of extension in the range of 50%. This study confirmed that cleaning using lasers with pulse durations in the pico- and femto-second range results into minimized alterations in the surface morphology and discolouration of the silver, compared to the use of nanosecond pulses. Moreover, in the case of the gilded fringes it was shown that it was easier to assess when the fringe was clean, as the un-corroded gold layer acts as a limit, given its compact surface and high reflectivity. From the above it is obvious that the decision on the optimum laser parameters in order to remove tarnish from silver and gilt silver threads with silk

ACKNOWLEDGEMENT This work was carried out at the Ultraviolet Laser Facility operating at IESL-FORTH with support from the access activities of the EC FP7Infrastructures-2007 project “Laserlab- Europe II” (Grant Agreement No: 212025). The authors are grateful to Annemette B. Scharff at the Royal Danish Academy of Fine Arts, School of

72

LACONA8.indb 72

11/10/2010 5:36:25 PM

Lerber, K. von, et al. 2005. Laser cleaning of silk: a first evaluation. ICOM Committee for Conservation 2: 978–988. Panzner, M., Wiedemann, G., Meier, M., Conrad, W., Kempe, A. & Hutsch, T. 2005. Laser cleaning of gildings, in Lasers in the conservation of artworks, LACONA VI, Springer proceedings in physics 116: 21–28. Pouli, P., Paun, I.-A., Bounos, G., Georgiou, S. & Fotakis, C. 2008. The potential of UV femtosecond laser ablation for varnish removal in the restoration of painted works of art. Applied Surface Science 254: 6875–6879. Reichert, U. 1998. Reinigungsversuge an Textilien mittels Lasertechnik- erste erfarungen. Restauro no. 6: 416–420. Siatou, A., Charalambous, D., Argyropoulos, V. & Pouli, P. 2006. A comprehensive study for the laser cleaning of corrosion layers due to environmental pollution for metal objects of Cultural value: preliminary studies on artificially corroded coupons. Laser Chemistry. Article ID 85324. Taarnskov, B. Laser cleaning of silver and gilt silver threads with silk core. Diploma thesis, Royal Academy of Art, School of Conservation, Copenhagen, 2009. Turovets, I., Maggen, M. & Lewis, A. 1998. Cleaning of Daguerreotypes with an Excimer Laser, Studies in Conservation 43: 89–100. Wise, E.M. & Coxe, C.D. 1961. Properties of precious metals: Silver and Silver Alloys. In Lyman, T. et al. (eds.), Metals Handbook (1), American Society for Metals: 1181–1189.

Conservation and to Yvonne Shashoua, The National Museum of Denmark, for help and advice. REFERENCES Baltova, S. & Vassileva, V. 1998. Photochemical behaviour of natural silk—II. Mechanism of fibroin photodestruction, Polymer Degradation and stability 60: 61–65. Belli, R. et al. 2006. Laser cleaning of artificially aged textiles. Applied Physics, A 83, DOI: 10.1007/00339006-3530-3. Bennett, J.M., Stanford, J.L. & Ashley, E.J. 1970. Optical Constants of silver sulfide tarnish films, Journal of the optical society of America, 60 (2): 224–232. Burmester, T., Meier, M., Haferkamp, H., Barcikowski, J., Bunte, J. & Ostendorf, A. 2005. Femtosecond laser cleaning of metallic cultural heritage and antique artworks. In LACONA V Proceedings, Springer Proceedings in Physics. (100): 61–69. CEI/IEC International Standard 68-2-46: Basic environmental testing procedures. Part 2: Tests. Guiding to test Kd: Hydrogen sulphide test for contacts and connections. IEC Publications 1982. Degrigny, C., Tanguy, E., Le Gall, R., Zafiropulos, V. & Marakis, G. 2003. Laser cleaning of tarnished silver and copper threads in museum textiles, Journal of Cultural Heritage 4: 152 s–156 s. Fotakis, C., Anglos, D., Zafiropulos, V., Georgiou, S. & Tornari, V. 2006. Lasers in the Preservation of cultural heritage. New York: Taylor and Francis. Lee, J.-M., Yu, J.-E. & Koh, Y.-S. 2003. Experimental study on the effect of wavelength in the laser cleaning of silver threads, Journal of Cultural Heritage 4: 157 s–161 s.

73

LACONA8.indb 73

11/10/2010 5:36:26 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Thickness of ablation control by structured light method R. Sitnik & J. Rutkiewicz Warsaw University of Technology, Warsaw, Poland

J. Marczak Military University of Technology, Warsaw, Poland

ABSTRACT: In the paper a novel on-line method for ablation thickness measurement of stone objects is proposed. Till now the process has been performed using knowledge and intuition of an experienced conservator. The proposed method belongs to the group of structured light techniques and is based on sinusoidal fringes and Gray code projection. The measurement process consists of the projection of a sequence of raster images on the surface of the object. After each projection the detector acquires an image representing the raster deformed by the object surface. 3D shape of the analyzed surface is calculated on the basis of captured images. Accuracy of the method is variable and mainly depends on the size of the measurement volume. To test the proposed method a limestone cross from Pauline Fathers monastery in Jasna Góra, Poland was cleaned by ReNOVALaser2 with varying fluencies. The whole process was monitored on-line by a structured light measurement system working with sensitivity below 1 μm. 1

INTRODUCTION

in journals without the possibility of real interdisciplinary debates and experience exchanges on the topic. Nowadays, application of laser technique gives the possibility of almost full control of the encrustation removal process at the surface of works of art. Selective and precise interaction of the light beam is the fundamental advantage of a non-invasive treatment of more or less tightly packed unwanted surface layers. Laser cleaning is becoming more and more popular when it comes to dirt removal. This method is non-contact, selective, local, controllable, self-limiting, gives immediate feedback and preserves even the gentlest of reliefs—the trace of a paintbrush (Salimbeni et al. 2000). At present the thickness of varnish can be controlled on-line (Gora et al. 2006). On-line controlling of ablation thickness of stone objects still remains a challenge. In this paper a structured light method (Sitnik et al. 2002) is proposed. Its main features are:

The idea of selective removing of optically absorbing matter (encrustation, soiled spots, etc.) from reflective surfaces with the aid of laser radiation is not a new one. It was demonstrated in 1965 by A. Schawlow (Schawlow 1965), one of the first authors of laser designs, utilizing an arrangement called “laser eraser”, selectively evaporating black absorbing pigments of printing ink from a strongly reflecting white paper sheet. At the beginning of the seventies of the previous century, J. Asmus and co-workers (Asmus et al. 1973) applied similar principle, utilizing ruby laser for removal of encrustation from white marble. Strongly absorbing black, lumped encrustation was removed with the aid of several laser pulses, while the cleaned surface of white marble remained intact, reflecting laser radiation. This novel approach did not overcome the experimental stage for several years, mainly because of the technological limit of pulsed laser sources available at that time. The situation changed during the nineties, mainly thanks to the stimulus provided to the research of innovative technologies dedicated to the study and safeguard of cultural heritage by European Framework Programs and various National Innovation Programs. Since then, the introduction of laser based cleaning techniques into conservation has led to a significant advance in the quality of cleaning possible for cultural artifacts. Up to 1995 the scientific results were reported on some conservation meetings, disciplinary congresses, and

• simplicity of setup in indoor and outdoor conditions, • calibration can be performed directly before measurements and requires no specialized equipment or knowledge, • the size of measurement volume can be adjusted according to required needs, • measurement uncertainty depends on measurement volume size (typically it is less than 1/10 000 with respect to volume size).

75

LACONA8.indb 75

11/10/2010 5:36:26 PM

2

The detector captures images of the calibration model in each position. From these images the relation between pixel co-ordinates (i, j) of the detector matrix and line formulas (A, B, C, x0, y0, z0) is calculated during an iterative process (see Figure 3). The process first calculates real model transformations for each position of the model and further line formulas in relation to detector co-ordinates. The algorithm is based on the minimization of least square distance between the imaging rays and (x, y, z) co-ordinates of local model positions. In the next step the user places the calibration model in phase positions (from 1CF to 4CF). For each position the whole sine and Gray codes sequence is captured. Then, on the basis of values obtained during the geometrical part of the calibration, the exact transformation of each phase model position is calculated. Thus, a sampled relation between phase and real (x, y, z) co-ordinates is calculated by using least squares method

STRUCTURED LIGHT METHOD

The proposed measurement method belongs to the structured light techniques group. Typical measurement setup is presented in Figure 1. It consists of a projector and a detector fixed together on a rigid frame. The common area of their fields of view forms the measurement volume. The method used is based on temporal phase shifting (Schwider et al. 1983) combined with hierarchical unwrapping (Osten et al. 1996). Temporal phase shifting is realized by the projection of a set of sinusoidal modulated patterns shifted in phase on the object surface. These patterns deformed on the object surface are acquired by the detector. After the projection of sine patterns the Gray code sequence is displayed. It realizes binary numbering of each period of the sine pattern. From the whole acquired sequence a phase map is calculated. It is further processed in order to obtain a set of (x, y, z) co-ordinates (cloud of points) representing sampled surface of the measured subject. For the calculation of (x, y, z) co-ordinates a calibration matrix is used. 2.1

Calibration

The goal of the calibration process is to find a method for scaling of phase values into real (x, y, z) co-ordinates. The proposed calibration method is an experimental one (Sitnik, 2005). It requires only a single known calibration model and allows to calibrate any volume that can be achieved by the hardware used. Calibration process is performed within two steps: geometrical and phase scaling. The only requirement for the user is to position the calibration model with an accuracy of several millimeters according to image presented in Figure 2. In the geometrical part the user places the calibration model in positions from 1CG to 6CG.

Figure 1.

Figure 2. Visualization of model positions which are required during the calibration process (CG—geometrical positions, CF—phase positions).

Figure 3. Visualization of model positions during the geometrical part of the calibration process.

Scheme of a typical structured light system.

76

LACONA8.indb 76

11/10/2010 5:36:26 PM

• time of a single measurement: 40 s, • number of points from a single measurement: 5,5 million.

optimization. Its representation is presented in Equation 1: F F k F k −l ⎧ k l m ⎪x p ∑ ∑ ∑ axklm l i j Φ ; k = 0 l = 0 m = 0 ⎪ F F − k F k −l ⎪ k l m ⎨ y p ∑ ∑ ∑ ayklm l i j Φ ; k =0 l =0 m=0 ⎪ F F − k F − k −l ⎪ k l m ⎪ z p = ∑ ∑ ∑ azklm l i j Φ ; k =0 l =0 m=0 ⎩

The whole calibration procedure was completed within one hour using an initially mounted setup. The computer used for calculation was a standard PC laptop with 4GB of RAM and Core2 Duo processor. The ablation has been performed by a Re NOVALaser2 system.

(1)

where axklm—best fitted polynomial coefficients; i, j—detector co-ordinates; Φ—phase value, F— order of best fitted polynomial; xp, yp, zp—resulting co-ordinates of subject surface. During the calibration process axklm and F are calculated. The final measurement uncertainty in the proposed method depends only on the calibration model accuracy and the resolutions of the projector and the detector used in the setup. 3

4

EXPERIMENT

As an exemplary subject a fragment of a limestone cross from Pauline Fathers monastery from Jasna Góra in Poland was selected (see Figure 5). This object has been kept outdoors for several decades now. Its whole surface is covered by encrustations. On the object surface four areas were selected for testing. Area A1 was consecutively irradiated with three different fluencies f1, f2 and f3 (corresponding to three laser pumping values equal to 30J, 35J and 42,5J). Areas A2, A3 and A4 were irradiated with single fluencies, respectively f3, f2 and f1. Exemplary visualization of shape differences in Area A1 during the experiment are presented in Figures 6, 7 and 8. To represent shape difference in the whole area of cleaning the Root Mean Square (RMS) value has been chosen. The value is calculated using distances between measured points before and after ablation in the normal direction to the local surface. For the A1 area the RMS values corresponding to fluencies are RMSA1f1 = 8 μm, RMSA1f2 = 20 μm and RMSA1f1 = 45 μm. For the areas A2, A3 and A4, which were irradiated with single fluencies

MEASUREMENT SETUP

To perform a validation of usage of the proposed structured light method in on-line monitoring of ablation process of stone objects, a system presented in Figure 4 was constructed. It consists of a DLP projector, a CCD detector and a fixing frame that rigidly connects these elements and the object plane. The DLP projector is based on the LCOS technology and its resolution is equal to 1920 × 1080 pixels. The CCD detector is a digital camera with the resolution of 3504 × 2336 pixels. The parameters of the presented system are: • measurement volume size: 50 mm × 50 mm × 30 mm, • measurement uncertainty: <2 μm,

Figure 5. Fragment of a limestone cross from Pauline Fathers monastery in Jasna Góra, Poland with four cleaned areas shown.

Figure 4. Photograph of the measurement setup (arrow visualizes the direction of ablation).

77

LACONA8.indb 77

11/10/2010 5:36:26 PM

Additionally the surface was very rough and heterogeneous, so the resulting ablation thickness can vary depending on local surface features. Presented results are very preliminary and in the future more complex measurements are planned together with statistical analysis for different materials. 5

Figure 6. tion (f1).

CONCLUSIONS

In the paper a new method for on-line monitoring of ablation process has been presented together with exemplary results. It is worth to mention that the proposed measurement method, including calibration in out-of-laboratory conditions is very easy to use. It requires no additional precise equipment apart from a calibration model. During experiment it has been proved that the developed method can be used for on-line monitoring of the ablation process. In the future the measurement system will be further improved with special focus on speeding up the measurement process in place of low measurement uncertainty.

Visualization of the A1 area after first abla-

ACKNOWLEDGEMENTS This work was performed under the grants No. R17 001 02 and N519 037 32/4193 financed by the Polish Ministry of Science and Higher Education. REFERENCES

Figure 7. Visualization of the A1 area after second ablation (f2).

Asmus, J.F., Murphy, C.G. & Munk, W.H. 1973. Studies on the interaction of laser radiation with art artifacts, Proceedings of SPIE 41: 19–27. Góra, M., Targowski, P., Rycyk, A. & Marczak, M. 2006. Varnish Ablation Control by Optical Coherence Tomography. Laser Chem. 2006: 10647. Salimbeni, R., Pini, R., Siano, S. & Calcagno, G. 2000. Assessment of the state of conservation of stone artworks after laser cleaning: comparison with conventional cleaning results on a two-decade follow up. J. Cult. Herit. 1: 385–391. Schawlow, A.L. 1965. Lasers, Science, 149: 13–22. Schwider, J., Burow, R., Elssner, K.E., Grzanna, J., Spolaczyk, R. & Merkel, K. 1983. Digital wave-front measuring interferometry: some systematic errors sources. Appl. Opt. 22: 3421–3432. Sitnik, R., Kujawińska M. & Woźnicki J. 2002. Digital Fringe Projection System for Large Volume 360-deg Shape Measurement. Opt. Eng. 41: 443–449. Sitnik, R. 2005. New method of structure light measurement system calibration based on adaptive and effective evaluation of 3D-phase distribution. Proc. SPIE 5856: 109–117. Osten, W., Nadeborn, W. & Andrae, P. 1996. General hierarchical approach in absolute phase measurement. Proc. SPIE 2860.

Figure 8. Visualization of the A1 area after last ablation (f3).

the RMS values are equal to RMSA2f3 = 16 μm, RMSA3f2 = 10 μm and RMSA4f1 = 8 μm. It is worth to mention that the ablation direction is not perpendicular to the surface which causes only some areas to be removed. Average angle of ablation was estimated to be around 45°.

78

LACONA8.indb 78

11/10/2010 5:36:29 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

213 nm and 532 nm solid state laser treatment of biogenetical fibrous materials M. Forster, S. Arif, C. Huber & W. Kautek Department of Physical Chemistry, University of Vienna, Vienna, Austria

S. Bushuk, A. Kouzmouk, H. Tatur & S. Batishche National Academy of Sciences of the Republic of Belarus, Institute of Physics, Minsk, Belarus

ABSTRACT: The preservation of the paper substrates by the minimization of the penetration depth of light in relation to a limited heat–affected zone is discussed. Potentials of this new technique are correlated with approaches with visible wavelength treatments at a model contamination consisting of charcoal powder on pure cellulose paper. The evaluation of cleaning thresholds, modification or ablation zones on the basis of microscopical diameter determinations was successfully applied to paper substrates the first time. The employment of far-UV laser radiation at 213 nm led to a lowering of the destruction threshold of paper due to the photochemical action of high energy photons in contrast to visible photons (532 nm) which result in thermal effects. A marked higher threshold was observed for 213 nm with the contaminant present due to the shading of the charcoal particles. Colorimetric lightness measurements served to quantify the cleaning status and the irreversible (photo) chemical modifications by measurement of yellowing. Yellowing and therefore chemical alterations have been observed at all optimum cleaning conditions at 532 nm in contrast to 213 nm. A cleaning mechanism has been detected for 532 nm radiation in addition to the well-known evaporation of contaminants, which is lateral mechanical blasting. 1

INTRODUCTION

has been recently demonstrated for laser cleaning of ancient leather (Batishche et al. 2007). There was the experimental indication that UV laser treatment does not lead to colorimetric alterations at 355 nm in analogy to visible radiation at 532 nm in contrast to 1064 nm [Rudolph et al. 2004, Kautek 2008]. In this new study, the extension to 213 nm radiation in comparison to 532 nm has been applied to a model paper such as additive-free cotton linters cellulose paper with no fillers and no sizing together with charcoal as a model contaminant.

Laser cleaning of fibrous materials can be successful when the interaction with the substrate is minimum based on a high penetration depth of laser light, which is the case with visible laser wavelengths (Kautek 2008, Kautek 2010). Numerous case studies demonstrated successful cleaning approaches even with complex pigmented objects [Kautek & Pentzien 2005, Holle et al. 2009]. An alternative approach can be a minimized penetration depth with far-UV radiation. Then, the laser can be used as ultra-precise non-contact scalpel. Thus, laser cleaning of stains and foxing from old paper and parchment may be much more effective and safe in contrast to visible and IR wavelengths where yellowing reactions may occur as an unacceptable side effect. There is the possibility to apply either ArF excimer (193 nm) or fluorine lasers (157 nm) (Sarantopoulou et al. 2003) which show however the disadvantage of aggressive gas supplies and/or the need of vacuum beam lines and vacuum working chambers, respectively. Therefore, the generation of the fifths harmonic of a Q-switched Nd:YAG laser appears as the most practical strategy. This technology provides also the option of intelligent wavelength mixing which

2

EXPERIMENTAL

2.1 Model system paper and charcoal contamination A model contamination was designed in order to study a well defined and reproducible system. Additive-free cotton linters cellulose paper without fillers and sizing was employed as substrate. Abraded charcoal was applied on the substrate, and loose remnants were brushed away. A conventional cleaning procedure was then applied before the laser cleaning process which consisted of brushing and vacuum cleaning. This final status would resemble

79

LACONA8.indb 79

11/10/2010 5:36:33 PM

that before a conventional rubber treatment which in contrast would lead to partial destruction of the paper surface.

translation razor blade x

laser beam

2.2

lens

A prototype Nd:YAG laser system with a powerful UV output of the 5th harmonic (213 nm) with up to 100 mJ, a repetition rate of 1–7 Hz and a pulse duration of 16 ns, served as the far-UV source (Fig. 1) The 2nd harmonic (532 nm) was delivered by a Nd:YAG laser (Quantel, Brilliant-EaZy) with a pulse frequency of 1–10 Hz equipped with a second harmonic module for the deliverance of up to 165 mJ at 532 nm. The beam was delivered via an attenuation optic consisting of a lambda-half plate and a polarizer. It was vertically focused by a plano-convex lens with a focal length of 500 mm onto a computer-controlled x-y stage for accepting the paper samples and a cutting blade serving for the precise determination of the beam diameters in the sample plane. The stage level was designed such that power meters could be placed below the stage. The determination of the beam diameter and crossection is of overruling importance in order to determine absolute correct fluence data. The used method consisted in the progressive shading of the beam during energy measurements in the x and y direction and the transmission energy measurements ET as a function of the stage coordinate x (Fig. 2) [Krüger & Kautek 2004]: ET ( x

detector

Laser cleaning systems

l)

∞ ⎡ ⎛ x ⎞2⎤ π F0 ⋅ ⋅ w0 ⋅ ∫ exp ⎢ −2 ⎜ ⎟ ⎥ dx 2 ⎢ ⎝ w0 ⎠ ⎥ l ⎣ ⎦

Figure 2. Razor blade technique for determination of beam radius [Krüger & Kautek 2004].

Figure 3. Representative transmission energy (ET) result of the razor blade technique according to Figure 2.

With a radius accuracy of ca. 10% this still resulted in area and fluence inaccuracies of 30%. A representative ET vs. x plot is depicted in Figure 3.

(1) 2.3

Diagnostics

In this context, the evaluation of cleaning thresholds and modification or ablation (destruction) zones on the basis of microscopical diameter evaluations were successfully applied to the paper substrates for the first time. In this technique, the diameter of the modified zones D is plotted versus the logarithm of the pulse energy E0 yielding the Gauss radius w0 (1/e2) from the slope and the threshold energy Eth from the abscissa intercept (Krüger & Kautek 2004): D2

Figure 1. Far-UV prototype laser system based on a 3-stage amplifier Nd:YAG laser with a powerful output of the 5th harmonic (5ω = 213 nm) with up to 100 mJ. 1—total reflection mirror; 2—LiF:F2− passive Q-switch; 3—diaphragm; 4–thin-film polarizer; 5, 13—active elements; 6—output resonant reflector; 7, 17—flash lamps; 8—deflecting mirrors; 9, 11—expanding telescopes with total magnification of 10x; 10—1x Keplerian telescope; 12—optical wedge; 14—λ/4 element; 15—focusing lens; 16—cell filled with SBSliquid; 18—KTP crystal; 19—DKDP crystal; 20—BBO crystal; 21—dispersing prism; 22—quartz plate; 23—glass plate; 24—energy meters.

⎛E ⎞ 2w02 ln ⎜ 0 ⎟ ⎝ Eth ⎠

(2)

Thus precise maximum fluence values in the Gaussian profile, F0, (Fig. 4) could be calculated from E0: F0 =

2 E0 π w02

(3)

Area values from this approach were in good agreement with the above-described razor blade

80

LACONA8.indb 80

11/10/2010 5:36:33 PM

Fluence F0

Fth F0/e2 -w0

0

w0

x Figure 5. Photographs with enhanced focus depth by reconstructed z-scans of the paper substrates (a) contaminated with charcoal, (b) laser-cleaned with 213 nm, N = 20, F = 0.16 J/cm2 (c) laser-cleaned with 532 nm, N = 9, F = 0.29 J/cm2.

Figure 4. Modification threshold evaluation on the basis of a Gauss beam profile. 2w0 represents the 1/e2-Gaussian beam diameter [Krüger & Kautek 2004].

method. This approach has been applied to many inorganic and polymer applications and proved also successful in this context. Colorimetry allows a relative, not absolute, comparison in respect to both lightness changes ΔL and saturation and hue changes given by the chromaticity coordinates Δa* and Δb* according to CIE-L*a*b* colour coordinates. Lightness changes ΔL and Δb*, describing the blue-yellow axis, could be correlated with the cleaning status. A hand-held spectrophotometer with a spectral resolution of 3.3 nm (SpectroEye, X-Rite Europe AG) was used. 3

RESULTS

3.1

Figure 6. D2-lnE plot. Cleaning (crossed symbols) and destruction results of clean (open symbols) and of charcoal contaminated paper (filled symbols). Triangular symbols: N = 5. Round symbols: N = 9.

Microscopic evaluation

Cleaning efficiencies and modification zones have been evaluated according to the procedures described in chapter 2.3. Figure 5 demonstrates an example of the sample status before and after cleaning steps. The initial contamination status shows that the carbonaceous particles are relatively inhomogeneously distributed on and between the fibres (Fig. 5a). This is also typical for the real contamination case. Laser treatment with the wavelengths of 213 nm (Fig. 5b) and 532 nm (Fig. 5c) resulted in surfaces almost comparable with the original. 3.2

The cleaning and destruction thresholds at 532 nm are strongly dependent on the number of pulses (Fig. 7). N = 1 leads to a cleaning threshold Fcl ≈ 1.5 J/cm2 which lies ca. 1.5 J/cm2 below the threshold of destruction in presence of the contaminant, Fth,c, which results in a “cleaning window” of 1.5 J/cm2. With N = 9, Fcl is 0.3 J/cm2 allowing a cleaning window of ca. 0.7 J/cm2 below the threshold of destruction in presence of the contaminant, Fth,c ≈ 1.0 J/cm2. Cleaning thresholds with 532 nm and N = 100 for charcoal treated pure cellulose paper (Whatman) which should be comparable with our additive-free cotton linters cellulose paper are reported for naked eye and colorimetric brightness inspection of the order of Fcl = 0.01–0.02 J/cm2 (Pentzien et al. 2010). These conditions of N = 100 lead to an even lower Fcl. However, in practice, this would not lead to an improved cleaning window, and further, a lower number of pulses would be advantageous for economical reasons. In that cited study (Pentzien et al. 2010), SEM inspection led to a direct determination of damage

Threshold determination

Representative threshold energy evaluations for cleaning and destruction with and without contaminant according to the D2-lnE relationship (Chapt. 2.3, Eq. 2) is shown in Figure 6. A constant slope indicates a reproducible focusing alignment and focus cross-section. Thresholds, both of cleaning and paper destruction, decrease with increasing number of pulses (Figs. 6 and 7), which is a direct indication of the phenomenon “incubation” (Krüger & Kautek 2004).

81

LACONA8.indb 81

11/10/2010 5:36:35 PM

Figure 8. Thresholds vs. pulse number at 213 nm. Destruction of the original paper (open triangles). Contaminated paper destruction (filled squares). Cleaning of contaminated paper (crossed circles).

Figure 7. Thresholds vs. pulse number at 532 nm. Contaminated paper destruction (filled squares). Destruction of original paper (open triangles). Cleaning of contaminated paper (crossed circles). For comparison: destruction of original paper with 213 nm.

thresholds for N = 10 varying between 11 J/cm2 for pure cellulose paper and around 2 J/cm2 for other paper types such as e.g. rag papers almost independent of the artificial aging status. Different degrees of polymerization could not explain the threshold trend. The destruction threshold for clean paper evaluated by the D2-lnE relationship in our present study yielded a value of Fth ≈ 1.3 J/cm2 for N = 9, which lies significantly below the above mentioned SEM evaluations. The employment of far-UV laser radiation of 213 nm led to a lowering of the destruction threshold of the pure paper, Fth, by about 50% in the range between N = 1 and 10 (Fig. 8). A marked higher threshold, Fth,c, was observed with the contaminant present. This surprising result may be explained by the shading of the charcoal particles in the N-range up to 60. Obviously, the penetration depth per pulse and therefore the ablation depth are so low that it takes up to 60 pulses to remove the particles completely. 3.3

Figure 9. Contaminated paper sample for colorimetric evaluation at 532 nm. Each parameter field separated by a mask contains laser spot arrays without overlap. Horizontal parameter is fluence (F0 = 0.13, 0.29, 0.48, 0.77, 1.92, 2.40, 4.80 J/cm2 from left) and the vertical parameter is the pulse number (N = 1, 2, 5, 9 from top down).

Δbmeasured (measured yellowing difference to clean status) and the contaminated status in respect to the uncontaminated status, ΔLi and Δbi, can be related to the actual value changes in respect to the uncontaminated status, ΔL (“cleaning”) and Δb, by

Colorimetric evaluation

ΔL =

Sample treatment for large areas necessary for the colorimetric evaluation was chosen in such a way for the first time that overlapping of spots could be avoided by a matrix treatment method depicted in Figure 9 where parameter fields (F, N) consisted of spot arrays separated with a periodicity of 1 mm. Therefore, the colorimetric data of these fields with a total area of Atotal do not represent directly the changes of the illuminated area ALaser, because there are unaffected areas, ABlack in between. The measured colorimetric values in respect to the uncontaminated status, ΔLmeasured (measured lightness difference to clean status) and

Δ

measured

Atotal ABlack ΔLi , ALaser

(4)

and Δb =

Δbmeasured Atotal ABlack Δbi . ALaser

(5)

The lightness results, ΔL, generally represent the cleaning status (Figs. 10 and 12), whereas the yellowing data Δb (Figs. 11 and 13) represent irreversible chemical and/or photochemical changes. All relevant cleaning data for 532 nm are contained in Fig. 10. Complete cleaning is represented

82

LACONA8.indb 82

11/10/2010 5:36:36 PM

Figure 10. Colorimetric evaluation for 532 nm treatment. Lightness, ΔL, values. White columns showing ΔL above Fth,c, grey columns below Fth,c.

Figure 12. Colorimetric evaluation for 213 nm treatment. Lightness, ΔL, values.

Figure 11. Colorimetric evaluation for 532 nm treatment. Δb values (positive yellow, negative blue). White columns showing yellowing above Fth,c, grey columns below Fth,c.

Figure 13. Colorimetric evaluation for 213 nm treatment. Δb values (positive yellow, negative blue).

by ΔL values approaching zero. This is the case for F0 = 0.77 J/cm2 at N = 1–9. Positive ΔL values above this fluence value are theoretically impossible. They can, however, be explained by a lateral mechanical blasting effect around the illuminated spots due to the expansion of vapour. This may originate either from the evaporation of the contaminant or of the paper being destructed. Thus, the test field e.g. with the parameters F0 = 1.92 J/cm2 and N = 5 (Fig. 9) shows a cleaning

condition also involving this blasting effect due to the evaporation of the contaminant. The test field beyond the destruction threshold e.g. with the parameters F0 = 4.80 J/cm2 and N = 9 show an almost completely cleaned status at the cost of substrate destruction and evaporation, respectively. The positive ΔL values above the destruction thresholds (Fig. 10) represent precisely this situation and are plotted white. All parameters leading to optimum cleaning conditions below the destruction threshold

83

LACONA8.indb 83

11/10/2010 5:36:38 PM

Table 1.

Cleaning and destruction parameters.

(N = 2)

532 nm

213 nm

Fc (optimum) Fc (cleaning threshold) Fth Fth,c

1.9 J/cm2 0.8 J/cm2 2.3 J/cm2 2.1 J/cm2

0.4 J/cm2 0.1 J/cm2 1.2 J/cm2 –

Yellowing and therefore chemical alterations have been observed at all optimum cleaning conditions at 532 nm. Yellowing obviously is caused by the bulk heating of the paper fibres by the deep penetration of 532 nm radiation, and not by the heat transfer from the evaporation plume of the contaminant. This is supported by the observation that 213 nm radiation can lead to evaporation of the contaminant without yellowing. With 213 nm treatment practically no blasting and yellowing effect occurred below the destruction threshold.

(see above) lead to finite positive Δb values and thus yellowing (Fig. 11). Δb is reduced again with increasing F0 and N due to the evaporative removal of the chemically modified paper material. With 213 nm treatment no blasting effect could be observed in the parameter range investigated (Fig. 12). The far-UV radiation causes a much smaller penetration depth of radiation in comparison to VIS radiation. Therefore, one has to realize that far-UV treatment will evaporate all materials illuminated, both contaminant and paper fibres, however in an extremely shallow range. Yellowing obviously is caused by the bulk heating of the paper fibres by the deep penetration of 532 nm radiation, and not by the heat transfer from the evaporation plume of the contaminant which either is just evaporated or even is reacting with air oxygen in a kind of burning process. This is supported by the observation that 213 nm radiation can lead to evaporation of the contaminant but not to yellowing (Fig. 13). Therefore, the heat transfer mechanism may be excluded. 4

ACKNOWLEDGEMENT We acknowledge partial financial support by the International Science and Technology Center, ISTC project B-1397. One of the authors (S.A.) thanks for a scholarship by the Higher Education Commission (HEC), Pakistan. REFERENCES Batishche, S., Kouzmouk, A., Tatur, H., Gorovets, T., Pilipenka, U., Ukhau, V. & Kautek, W. 2007. Simultaneous UV-IR Nd:AYG Laser Cleaning of Leather Artifacts. In Lasers in the Conservation of Artworks, Springer Proceedings in Physics 116: 221. Holle, H., Kautek, W., Pentzien, S., Krüger, J., Mäder, M. & Schreiner, M. 2009. Laser Cleaning on Historic Picture Postcards. In Engel P. (ed.), Research in Book and Paper Conservation in Europe—a State of the Art, Verlag Berger: 189. Kautek, W. & Pentzien, S. 2005. Laser Cleaning System for Automated Paper and Parchment Cleaning. Springer Proceedings in Physics, Vol. 100, (eds.) Dickmann, K., Fotakis, C. & Asmus, J.F.: 403. Kautek, W. 2008. Laser Cleaning of Paper and Other Organic Materials. In M. Schreiner, M., Strlič. & R. Salimbeni (eds.). Handbook on the Use of Lasers in Conservation and Conservation Science: Chapt. 2.4. COST office, Brussels 2008. http://www. science4heritage.org/COSTG7/booklet/ Kautek, W. 2010. Lasers in Cultural Heritage: The NonContact Intervention. In Miotello, A. & Ossi, P.M. (eds.), Laser-Surface Interactions for New Materials Production: Tailoring Structure and Properties, Springer Series in Materials Science 130: 313. Krüger, J. & Kautek, W. 2004. Ultrashort Pulse Laser Interaction with Dielectrics and Polymers. In Advances in Polymer Science, Vol. 168, (ed.) Lippert, T., Springer-Verlag Berlin Heidelberg: 247. Pentzien, S, Conradi, A. & Krüger, J. 2010. This volume. Rudolph, P., Ligterink, F.J., Pedersoli, Jr., J.L. & van Bommel, M., Bos, J., Aziz, H.A., Havermans, J.B.G.A., Scholten, H., Schipper, D. & Kautek, W. 2004. Characterization of laser-treated paper. Appl. Phys. A 79: 181. Sarantopoulou, E., Samardzija, Z., Kobe, S., Kollia, Z. & Cefalas, A.C. 2003. Removing foxing stains from old paper at 157 nm. Appl. Surf. Sci. 208–209: 311.

CONCLUSION

Both cleaning and paper destruction thresholds decrease with increasing number of pulses due to incubation. The most relevant cleaning and destruction parameters are collected in Table 1. The employment of far-UV laser radiation of 213 nm led to a lowering of the destruction threshold of paper due to the photochemical action of high energy photons in contrast to visible photons (532 nm) resulting in thermal effects. A marked higher threshold, Fth,c, was observed for 213 nm with the contaminant present due to the shading of the charcoal particles. The lightness, ΔL, results generally represent the cleaning status, whereas the yellowing data Δb represent irreversible chemical and/or photochemical changes. A cleaning mechanism has been detected for 532 nm radiation in addition to the well-known evaporation of contaminants. Lateral mechanical blasting is able to remove material near the illuminated spots due to the expansion of contaminant vapour.

84

LACONA8.indb 84

11/10/2010 5:36:39 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Free-running Er:YAG laser cleaning of mural painting specimens treated with linseed oil, “beverone” and Paraloid B72 J. Striova & E. Castellucci Department of Chemistry, University of Florence, Italy

A. Sansonetti, M. Camaiti & M. Matteini ICVBC-CNR Istituto per la Conservazione e la Valorizzazione dei Beni Culturali, Florence, Italy

A. deCruz Department of Chemistry, Duke University, Durham, NC, US

A. Andreotti & M.P. Colombini Department of Chemistry and Industrial Chemistry, University of Pisa, Italy

ABSTRACT: In this paper, preliminary results of laser-assisted (free-running Er:YAG laser) removal of several organic materials (i.e. linseed oil, “beverone”, Paraloid B72) applied to the surfaces of laboratory samples of mural paintings, are reported. The specimens were prepared with “a fresco” technique, using Saint John’s white and yellow ochre pigments. The optimal cleaning conditions were established between the ablation threshold of the surface treatments and the damage threshold of the pictorial layers. The laboratory samples were subjected to laser ablation in order to establish optimal cleaning conditions which were as regards for example linseed oil: 0.6–1.3 J/cm2, in the presence of isopropanol and an eventual chemical pre-treatment. These surfaces were characterized morphologically, by colorimetric measurements and microRaman spectroscopy, all measurements carried out before and after laser radiation. Laboratory experimentation provided useful information for cleaning tests performed in situ and on the ancient fragments of the mural paintings from the Camposanto Monumentale in Pisa. 1

results obtained with first harmonic of the Nd:YAG laser (1064 nm). The laser cleaning with intermediate pulse duration (short free-running) (Andreotti et al. 2006, Siano et al. 2007) enabled controlled removal of various undesired layers and helped to solve complex conservation problems of wall painting cleaning even though encouraging results were achieved also in picosecond regime (Andreotti et al. 2006). Several research groups investigate the phenomenology and effects of the most diffused laser system (Nd:YAG) in the field of cultural heritage cleaning. A thorough examination of the interaction of other laser wavelenghts, such as that produced by the Er:YAG laser at 2.94 μm, with material involved in the craftsmanship and conservation of mural paintings is however absent. This fact motivates our study which presents the preliminary results in evaluation of potentiality of this laser system, operated in freerunning regime, in cleaning of mural paintings. Laser cleaning, in general, is based on the absorption of the laser radiation by the material to be ablated. Many organic substances do absorb at

INTRODUCTION

Cleaning of mural paintings is challenging and complex due to the artwork surface morphology and the chemical composition of the materials. Moreover, the paint layers on mural substrates have often been treated with both organic and inorganic fixatives. The ageing and deposition processes may cause the formation of a complex surface matrix strongly adherent to the substrate and to which traditional cleaning methods will not work for a satisfactory removal. More effective and less invasive solutions should be searched for each restoration intervention. Laser-assisted cleaning tests, reported in literature, on real mural paintings or laboratory models involved various wavelenghts accross the electromagnetic spectrum from ultraviolet, visible, near-infrared (Shekede, 1997, Gaetani et al. 2000, Siano et al. 2007, Andreotti et al. 2006) to mid-infrared radiation (Andreotti et al. 2007). Some studies investigated laser pulse width effects in cleaning of wall paintings (Andreotti et al. 2006, Siano et al. 2007) and reported on the promising

85

LACONA8.indb 85

11/10/2010 5:36:39 PM

2.94 μm due to the presence of OH groups in their moleculare structure. The choice of this wavelength in removal of organic substances may be justified because this radiation is in resonance with the vibrational energy of OH molecular bond. The process of ablation may be further enhanced by introducing the OH bearing liquids such as isopropanol or water that may increase absorption of the laser radiation. Another advantage of using the auxiliary -OH containing wetting agents may consist in confinement of surface temperature. (Vogel and Venugopalan 2003). One of the most important issues in the laser cleaning of the polychrome surfaces is the potential damaging effects of the laser radiation to pigments (Bordalo et al. 2006). In a previous study, the interaction of copper-based pigments (malachite and azurite) with the Er:YAG radiation was investigated (Camaiti et al. 2008), as a part of an extent survey aimed at the characteriazion of the laser-matter interaction associated with the 2.94 μm wavelenght. In this work, the sensitivity of Saint John’s white and yellow ochre pigments to laser radiation was explored. These two pigments were chosen because they were frequently used in paint layers of mural paintings. Saint John's white (CaCO3 formed from slaked lime) forms a durable pictorial layer when applied with “a fresco” technique (Cennini 1999), moreover the white surface displays well the potential chromatic changes induced by the laser radiation. This pigment is not expected to be sensitive to the 2.94 μm laser radiation because no -OH groups are present in its chemical composition, when the slaked lime completes the carbonation. Also, yellow ochre forms a durable pictorial layer when applied “a fresco” (Cennini 1999). However, it contains -OH groups (goethite—FeO(OH)) and therefore the pigment may be sensitive to the Er:YAG radiation. A lower damage/alteration threshold is expected for this pigment as compared to Saint John's white one. Laser-assisted removal of several organic materials (linseed oil, Paraloid B72, “beverone”) from the surfaces of these specimens is analyzed and reported. These organic treatments, among others, were also detected on the pictorial surfaces of Camposanto Monumentale in Pisa during recent analyses (Colombini 2008, Biondelli et al. 2008) and fully justify the choice of these materials in this experimentation. In fact, “beverone” (mixture of egg, rabbit skin glue, vinegar and water), linseed oil and Paraloid B72 were often used by conservators in past restoration treatments, with intent to revive colors and/or to consolidate the detached paint layers. All these treatments do absorb at 2.94 μm except for PB72. In the latter, the laser ablation is expected to be mediated by introduction of OH bearing wetting agent. The results obtaiend from

this study may contribute the current knowledge of this wavelenght interaction with materials inherent to mural painting and may promote the safe application of the Er:YAG laser radiation in cleaning of real mural paintings. 2

MATERIALS AND METHODS

2.1 Specimens preparation The pigments used were supplied by Zecchi, Florence and their purity was controlled by XRD analysis. In the yellow ochre formulation a certain amount of gypsum was detected, probably used as a thickener. Actually gypsum is often detected in commercial pigments and its presence has been exploited to argue further information about its interaction with used laser radiation. Specimens simulating mural paintings were prepared using a porous calcarenite substrate on which a plaster, composed of aerial lime and siliceous sand (1:2 in volume), was spread. The pictorial layer was applied “a fresco”, by mixing the pigment powder in water, on the freshly prepared plaster substrate. Linseed oil, “beverone” and Paraloid B72 (5 wt.% in acetone) were applied by brush on the sample surfaces to simulate past restoration interventions. 2.2 Laser device and irradiation procedure The laser tests were performed with an Er:YAG laser at 2.94 μm, High Power Erbium CrystaLase, (MonaLaser; LLC of Orlando, Fla). This laser emits a pulse width of 300 μs “macro-pulse” consisting of a train of 1–2 μs micro-pulses about 2 μs apart. The pulse from the High Power Erbium CrystaLase is delivered with an articulated arm. The tests were carried out with a pulse repetition rate of 10 Hz and a spot diameter of 1 mm. Damage thresholds The irradiation procedure consisted in irradiation of 25 cm2 surface of each specimen divided into 9 equal sectors. Individual areas were irradiated in a manual scanning manner to simulate typical cleaning process during the restoration intervention. This corresponds to approximatelly 10 pulses per spot. As mentioned before, the laser ablation by Er:YAG laser can be carried out in dry or wet conditions. For this purpose, three sectors (S1, S2 and S3) were treated in dry conditions with laser radiation of increasing fluence, in order to identify a damage threshold and then to observe the radiation effects and higher fluences. To examine the effects of laser irradiation in wet conditions, other three sectors (S4, S5 and S6) were irradiated with the same criteria using water as a

86

LACONA8.indb 86

11/10/2010 5:36:39 PM

that was kept below 0.6 mW in order not to induce any thermal damage to the studied material. MINOLTA CR 200 colorimeter was used to measure the surface color, using CIE L*a*b* system. Color measurements were carried out before and after irradiation with the aid of a grid in order to identify the identical micro-areas. Total color difference (ΔE) was calculated according to:

wetting agent. In fact, this wetting agent reaches absorption maximum at 2.94 μm (S. Georgescu and O. Toma 2005) and may enhance the absorption of the laser radiation. In the remaining three sectors (S7, S8 and S9), the effects of the laser irradiation in presence of isopropyl alcohol was investigated. This alcohol was chosen because of its good wetting properties, suitable boiling point (82°C) and volatily (33 mmHg at 20°C) as compared to other alcohols. Furthermore, the speed of alcohol evaporation was reduced by interposing a glass slide in between the sample and laser radiation path. This procedure also served to collect ablated material.

Δ

Δb*2

(1)

where ΔL* is the difference in lightness, while Δa* e Δb* represent the difference in chromaticity coordinates before and after irradiation.

Ablation thresholds and cleaning experiments 20–40 μm thin films, as measured by Electronic Digital micrometer Microstat MS25 (Cadar), were prepared by applying linseed oil (not diluted), PB72 (5% in acetone, 0.25 mL), egg white (not diluted), egg yolk (50% in water) on 2.5 × 1 cm microscope glass slides. The ablation thresholds were determined by applying a series of laser pulses at variable fluences (in sequence of 12.7, 6.3, 3.2, 1.9, 1.3, 0.6 J/cm2) in adjacent spots. Irradiation was performed in dry and then in wet conditions with frequency kept at 2 Hz. The laser assisted removal of the linseed oil, “beverone” and P-B72 treatments from St. John’s white and yellow ochre mural painting specimen was carried out. For this purpose, the specimen surfaces were divided into four sectors: sector 1 was utilized for preliminary tests; then sectors 2, 3 and 4 were used for alternative cleaning conditions which are described in the text (individual areas were irradiated in a manual scanning manner to simulate typical cleaning process during the restoration intervention). All the tests were concluded with cleaning finishing by soft brush moistened by corresponding wetting agent. 2.3

ΔL*2 + Δa*2

3

RESULTS AND DISCUSSION

3.1 Damage thresholds Irradiation of “a fresco” prepared specimen in dry conditions produced damage thresholds at around 15.2 J/cm2 for St. John’s white and 1.9 J/cm2 (300 pulses) and 2.5 (10 pulses) J/cm2 for the yellow ochre. As regard the white specimen, the plasma formation was observed in form of bright sparks (starting from 15.2 J/cm2). At this point, it has been observed that the material is removed. Plasma mediated removal increased and became more uniform with further increase in fluence. Relatively low alteration threshold of the yellow ochre specimen (starting from 1.9 J/cm2) is due to the partial transformation of the yellow ochre composed of goethite and gypsum into hematite and bassanite (as determined by Raman spectroscopy, spectra not shown) and to the uncovering of quartz crystals coming from the plaster aggregate fraction. The irradiation of surface material, moistened by water or alcohol, with the Er:YAG laser lead to micro-explosions of the wetting agent droplets that caused the ejection of the adjacent material at fluences starting from 2.5 and 3.8 J/cm2 for St. John’s white and yellow ochre specimen (isopropanol), respectively, and for both specimens at 2.5 J/cm2 (water). The higher removal rates of pictorial layer and more homogeneous ablation have been observed in the case of white specimen in the presence of water as compared to isopropanol. The reason could be found in different penetrations depth of the wetting agents. Isopropanol penetrates faster and deeper into the sample, due to its lower surface tension, and evaporates faster due to its higher vapor pressure. On the other hand, water penetrates less and evaporates slower, which may imply more effective laser-water interaction within the outer layers of white specimen. The yellow ochre specimen chromatic variations caused by irradiation at different fluences, in dry

Evaluation of laser irradiation effects

Optical microscopy in the visible and ultraviolet light was used to observe surface morphology and to follow the removal of the superficial organic treatments (Nikon Eclipse E600 Stereomicroscope equipped with a digital image capturing system). A series of microphotographs was acquired before and after the laser irradiation. Raman Renishaw System 2000 spectrometer coupled to Leica optical microscope, in confocal configuration, was used to evaluate the potential chemical alteration induced by the laser irradiation of the pure pigments or mural painting specimens. A 785 nm diode laser was used as an excitation source. A series of neutral density filters was used to attenuate the laser power arriving on the sample

87

LACONA8.indb 87

11/10/2010 5:36:39 PM

conditions, in presence of water or isopropanol, are shown in Table 1. ΔE becomes significant starting from 3.8 J/cm2 for irradiation in dry conditions, and from 5.1 J/cm2, as concerns the irradiation in presence of water and isopropanol. The increase of the damage threshold in the presence of the wetting agents was also verified by Raman measurements; however no compositional changes of the pigment constituents were evidenced. 3.2

applied in form of thin film, with laser radiation was first established (Table 2). The higher ablation threshold for linseed oil and egg yolk (dry conditions-6.3 J/cm2) as compared to egg white and rabbit glue is probably due to fatty components. However, when a small amount of isopropanol or water was introduced some disturbances were induced at 0.6 J/cm2, even in the egg yolk. Moreover, the chromatic variations in the thin films, caused by the thermal action of the laser radiation, observed in dry conditions, were limited in the presence of wetting agents.

Ablation thresholds and laser assisted cleaning of mural painting specimens

3.2.1 Linseed oil The preliminary cleaning tests carried out in specimen sector 1 indicated the optimal cleaning conditions at 1.3 J/cm2 (10 mJ/pulse) and isopropanol as wetting agent. Three alternative conditions of cleaning were performed in sectors 2, 3 and 4 as shown in Table 3 both for St. John’s white and yellow ochre specimen. The effects observed when using 0.6 and 1.3 J/cm2, sector 2 and 4, were quite similar in the final effects, but slightly different in intensity. Specifically, the softening of linseed oil, induced by laser irradiation in the presence of isopropanol, was achieved and the linseed oil residues were removed by gentle passage with a soft brush moistened with isopropanol. Chemical pre-treatment (sector 3, Table 3) was aimed on partial hydrolyzation of ester bonds of glycerol and fatty acids in order to facilitate the oil removal with successive laser cleaning at low fluence. Indeed, the oil layer was softened significantly by the poultice application and the satisfactory and relatively fast oil removal was achieved at 0.6 J/cm2 fluence, in the presence of isopropanol. The great chromatic variations (ΔE = 10.6) induced by linseed oil treatment on the white specimen, prevalently due to the increase of b* parameter (yellowing) (Figure 1), allowed for monitoring of treatment removal. In case of a complete oil removal, and without any other damage, the chromatic differences between the cleaned and original surface should be close to zero.

The optimal working conditions for removal of materials from the surfaces were established in the narrow equilibrium between the ablation thresholds of the surface layers (determined on thin films prepared on and shown in Table 2) and the damage threshold of the pictorial layers. In general, ablation thresholds were lower during irradiation in wet conditions. Ablation threshold of each component of the “beverone” treatment, Table 1. Chromatic variations, at different fluences (F), of yellow ochre specimens’ surfaces irradiated with the Er:YAG laser. ΔL*

Δa*

Δb*

ΔE

S1 S2 S3

1.0 1.1 0.5

−0.3 −0.8 −0.5

−1.9 −4.4 −6.2

2.2 4.6 6.2

1.3 3.8 6.3

Water S4 S5 S6

−0.4 −0.2 −2.0

0.0 −0.8 0.9

−1.3 −3.1 −5.0

1.4 3.2 5.5

2.5 5.1 11.4

iPr Al S7 S8 S9

1.1 0.5 −1.0

−0.6 −1.4 0.7

−2.9 −6.2 −8.9

3.2 6.4 9.0

2.5 5.1 8.9

Sector Dry

F[J/cm2]

Table 2. Ablation thresholds individuated on organic thin films expressed in fluence (F). Treatment

Ablation condition

Ablation fluence [J/cm2]

Linseed oil

Dry Isopropanol Dry Isopropanol Dry Water Dry Water Dry Water

6.3 0.6 – 0.6 1.3 0.6 1.3 0.6 6.3 1.3

PB72 Rabbit glue Egg white Egg yolk

Table 3. Conditions of cleaning tests employed for removal of oil treatment from the surface of mural painting specimens. Sector S2 S3

S4

Chemical pre-treatment

Wetting agent

Fluence [J/cm2]

– Poultice 5 wt% (NH4)2CO3 in H2O 10 min –

Isopropanol Isopropanol

1.3 0.6

Isopropanol

0.6

88

LACONA8.indb 88

11/10/2010 5:36:39 PM

Figure 1. Chromatic variations in St. John’s white specimen.

Figure 2. Raman spectra acquired on the yellow ochre specimen in (a) S2, (b) S3 (c) S4 sectors, after the laser assisted cleaning.

All the cleaned surfaces appeared less yellow (smaller Δb*) as compared to the surface treated with oil. This means that the laser irradiation did not cause yellowing intensification and the yellow oil treatment has been, at least partially, removed. From the chromatic point of view, the cleaning in sector 3 (combined chemical/physical cleaning) was the least harmful and the most efficient. Chromatic variation in S2 and S4 suggest less advanced cleaning as compared to S3. Raman spectra acquired on ochre specimen in S2, S3 and S4 after the laser assisted cleaning are shown in Figure 2. All the spectra exhibit signals typical for goethite (299, 386, 551 cm−1), gypsum (494, 1007 cm−1) and calcite (278, 712, 1085 cm−1). The sulphate bands of gypsum at 1007 cm−1 do not exhibit any broadening due to the co-presence of sulphate hemi-hydrate in any of the registered spectra, and no hematite bands are present. These data show the absence of any chemical variations in the pigments at the tested fluences. Moreover, from a morphological point of view the specimen texture and morphology have not changed.

Table 4. Conditions of cleaning tests employed for removal of “beverone” treatment from the surface of mural painting specimens. Sector S2 S3 S4

Chemical pre-treatment

Wetting agent

Fluence [J/cm2]

– 1% NH4OH in H2O 5 min –

Water 1% NH4OH in H2O Water

1.3 0.6 0.6

of the yellow ochre specimen have been slightly more intense in Sectors 2 and 4 (ΔE = 1.7, 0.6, 2.4, respectively for sectors 2, 3, 4). Morphological features of the painted layers were well preserved for all the cleaning conditions (example in Fig. 3a–b). An example: the effective removal of the “beverone” treatment is demonstrated by microphotographs of the UV-induced fluorescence, taken before and after laser irradiation. “Beverone” exhibits a very strong light-blue fluorescence (Fig. 3c), which completely disappeared after laser ablation (Fig. 3d).

3.2.2 “Beverone” On the basis of the preliminary cleaning tests and ablation thresholds three alternative cleaning conditions were used as described in Table 4. Water as wetting agent and mild irradiation conditions were used in S2 and S4. Water was used since it proved to be more powerful in preventing thermal deterioration of the irradiated materials, as compared to isopropanol. Cleaning effects, macroscopically observed in the three sectors were very similar. As regards St. John’s white specimen, the analysis of chromatic data indicates that the extent of cleaning is satisfactory for all the sectors (ΔE = 1.0, 1.2, 1.3 respectively for sectors 2, 3, 4). Chromatic changes between cleaned and original surfaces

3.2.3 “Paraloid B72” The removal of the acrylic polymer Paraloid B72 (PB72) was not achieved by the irradiation in dry conditions. This is due to the fact that there are no O-H groups in the structure. On the contrary, by using a small amount of a wetting agent, such as isopropanol, it’s possible to induce the ablation process. This moderate penetration into the polymer structure probably caused a photo-mechanical disruption of the polymer by the rapid expansion of the wetting agent. The damage threshold for St. John’s white and yellow ochre specimen was found at F = 2.5 J/cm2 (Falt), in presence of

89

LACONA8.indb 89

11/10/2010 5:36:40 PM

those obtained with acetone/isopropanol mixture at 0.6 J/cm2. 4

The pulsed laser at 2.94 μm (Er:YAG laser, 300 μs pulse) was tested as a cleaning tool in removal of several organic (linseed oil, “beverone”, Paraloid B72) treatments from a fresco prepared mural paintings specimens. Firstly, damage thresholds of specimens were found at 15.2 and 2.5 J/cm2 (10 pulses/spot) respectively for paint layers with St. John’s white and yellow ochre pigment, in dry irradiation conditions. This finding is in agreement with the expected results and confirms the lower damage threshold for pigment containing -OH groups. For the irradiation in wet conditions, the damage thresholds are comparable for both pigments. In specific a damage threshold of 2.5 and 3.8 J/cm2 was found for St. John’s white and yellow ochre paint layer in presence of water and 2.5 J/cm2 in presence of isopropanol for both paint layers. Ablation thresholds of components of organic treatments (linseed oil, PB72, rabbit glue, egg white, egg yolk) were individuated by test performed on respective thin films. Lower ablation thresholds were measured in presence of wetting agents, in specific, 0.6 J/cm2 for linseed oil and PB72 (both with isopropanol), for rabbit glue and egg white (both with water) and 1.3 J/cm2 for egg yolk (with water). Moreover, thermal effects of the laser radiation observed on the rabbit glue, egg white and yolk films in dry conditions were absent in presence of wetting agents. Ablation of PB72 was possible only in presence of isopropanol. Three alternative conditions with fluence contained in the interval between ablation and damage thresholds were tested for removal of each of three organic treatments. Laser ablation at 0.6 and 1.3 J/cm2 in presence of isopropanol, enabled a gradual removal of the oil layer, especially when combined with a soft mechanical action. A greater efficiency was achieved by combining the chemical treatment and the subsequent laser-assisted cleaning and would be the preferred choice for surfaces where ammonium presence is not dangerous. Efficient removal of the “beverone” treatment was achieved in all tested sectors, even in very mild radiation conditions (0.6 J/cm2), with water as wetting agent. Removal of Paraloid B72 was possible at 0.6 J/cm2 fluence in presence of isopropanol but required long cleaning time. Cleaning procedure with acetone/isopropanol mixture at this

Figure 3. St. John’s white specimen. S4 before (left) and after (right) cleaning in visible (top) and in fluorescenceinducing light (bottom). λex = 330, 380 nm; scale bar = 1 mm. Table 5. Conditions of cleaning tests employed for removal of Paraloid B72 treatment from the surface of mural painting specimens. Sector

Chemical pre-treatment

Wetting agent

Fluence [J/cm2]

S2 S3

– Acetone/ Isopropanol (1:1) - 1 min –

Isopropanol Acetone/ Isopropanol (1:1)

1.3 0.6

Isopropanol

0.6

S4

CONCLUSIONS

isopropanol; on the basis of these results working fluence (Fw) has been kept Fw < Falt as shown in Table 5. Chromatic impact of the Paraloid B72 treatment on the original surface was not significant (ΔE = 1.0); this means that chromatic measurements after laser irradiation may serve only for evaluation of the negative side effects. No significant changes were detected in chromatic measurements or morphological observations. The efficiency of the removal of the Paraloid B72 was evaluated by confocal micro-Raman spectroscopy that enabled estimating the thickness of the organic film on the specimen surface, which was around 8–10 μm before the laser removal. A detailed procedure of these measurements is reported by Striova, 2009. In brief, the P-B72 removal by the laser ablation at 0.6 J/cm2 in the presence of the acetone/isopropanol mixture (S3) proved to be more efficient then that in presence of pure isopropanol (S4), probably due to the partial solvent action of the acetone. The P-B72 ablation in the presence of isopropanol at a fluence of 1.3 J/cm2 (S2) provided similar results to

90

LACONA8.indb 90

11/10/2010 5:36:40 PM

2.94 μm with azurite and malachite pigments” in Proc. of Lacona VII, CRC Press, 2008, 253–258. Cennini, C. Il Libro dell’Arte. Serchi M. (ed). Felice Le Monnier. Firenze (1999). Colombini, P. Caratterizzazione dei materiali organici in I1 camposanto di Pisa. Opera Primaziale Pisana, 2008. Gaetani, M.C. and Santamaria, U. 2000, The laser cleaning of wall paintings, J. Cult. Herit. 1, S199–S207. Georgescu, S. and Toma, O., 2005Er:YAG Three-Micron laser: performances and limits. IEEE J Sel Top Quant, 11: 682–689. Shekede, L. 1997, Lasers: a preliminary study of their potential for the cleaning and uncovering of wall paintings ed. W. Kautek and E. König, Proc. of Lacona I, 51–56. Siano, S., Brunetto, A., Mencaglia, A., Guasparri, G., Scala, A., Droghino, F. and Bagnoli, A. 2007. Integration of laser ablation technique for cleaning the wall paintings of Sacrestia vecchia in S.M. della Scala. Proc. Lacona VI, 191–201. Striova, J. “The Er:YAG as a controlled tool for the cleaning of the mural paintings” PhD dissertation, University of Florence, 2009. Vogel, A. and Venugopalan, V. 2003, Mechanisms of pulsed laser ablation of tissue, Chem. Rev. 103, 577–644.

fluence increased the cleaning efficiency, due to the combined chemical and physical action, which was comparable to that achieved at 1.3 J/cm2, in presence of isopropanol. REFERENCES Andreotti, A., Colombini, M.P., Nevin, A., Melessanaki K., Pouli, P. and Fotakis, C. 2006, Laser Chemistry, vol. 2006, Article ID 39046. doi:10.1155/2006/39046. Andreotti, A., Colombini, M.P., Felici, A., deCruz, A., Lanterna, G., Nakahara, K. and Penaglia, F. 2007. Preliminary results of the Er:YAG laser cleaning of mural paintings. Proc. Lacona VI, 203–210. Biondelli, D., Camaiti, M., Matteini, M., Sansonetti, A., Zerbi, C., Castellucci, E. and Striova, J.: Caratterizzazione dei materiali organici e inorganici in Il Camposanto di Pisa. Opera Primaziale Pisana, 2008. Bordalo, R., Morais, P.J., Gouveia, H. and Young, C., 2006, “Laser Cleaning of Easel Paintings: An Overview,” Laser Chemistry, vol. 2006, Article ID 90279, doi:10.1155/2006/902792006. Camaiti, M., Matteini, M., Sansonetti, A., Striova, J., Castellucci, E., Andreotti, A., Colombini, P., de Cruz, A. and Palmer, R. “The interaction of laser radiation at

91

LACONA8.indb 91

11/10/2010 5:36:40 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Studies on the UV femtosecond ablation of polymers: Implications for the femtosecond laser cleaning of painted artworks I.A. Paun Faculty of Applied Sciences, University Politehnica of Bucharest, Bucharest, Romania

A. Selimis, G. Bounos & S. Georgiou Institute of Electronic Structure and Lasers (IESL), Foundation for Research and Technology-Hellas (FORTH), Heraklion, Crete, Greece

ABSTRACT: Due to its well known advantages, the use of femtosecond ( fs) laser technology promises the means for overcoming the limitations of nanosecond (ns) laser-based restoration of painted works of art. However, for a safe utilization on actual artworks, the effects induced in the substrates by fs laser pulses must be carefully examined. To this end, we investigate laser ablation of polymers with UV (248 nm) ultrashort (500 fs) laser pulses. For a simplified approach, we study polymers (Poly-(methyl methacrylate) PMMA) doped with aromatic compounds, systems that are chemically similar to the painting materials. A very important parameter for our study is polymer Molecular weight (Mw), which relates to the number of bonds that must be broken for material ejection to occur. Mw is important also in applications, since it can vary from layer to layer, or from one painting to another. We investigate the etching resolution, the morphology of the etched areas and the chemical modifications induced in the remaining substrate following laser-cleaning protocols. We find that the etching efficiencies and the morphological modifications upon UV fs irradiation differ drastically from the ones observed in the corresponding ns irradiation. Furthermore, we find a strong dependence of these effects on polymer MW. Possible mechanisms for polymer decomposition/ablation, as well as the implications for the laser-restoration performances are discussed. 1

INTRODUCTION

Furthermore, the lack of plasma shielding allows a maximum coupling of the laser energy into the substrate (Heitz et al. 1994). However, certain aspects of the interaction of fs laser pulses with materials are expected to differ substantially from the case in which longer pulse durations are employed. Despite the extensive interest, the mechanism of UV fs ablation of polymeric materials is not yet elucidated. In consequence, before femtosecond laser techniques are applied to real samples (works of art), various aspects of the interaction of fs laser pulses with these substrates must be carefully examined. For instance, due to the increased complexity and photolability of painted artworks, for a safe utilization of laser-cleaning techniques, detailed examination of the extent of morphological and chemical modifications induced in the irradiated substrates is strongly required (Küper et al. 1987, Heitz et al. 1994, Preuss et al. 1993). In order to meet these objectives, as well as for providing a framework for the development of safe restoration protocols, the present study addresses particular features of fs ablation of polymeric systems.

Laser ablation has evolved into a powerful tool for the restoration of painted artworks, offering significant advantages over the traditional techniques (Zafiropulos et al. 1998, Georgiou et al. 1998). For example, when removal of degraded varnish layers from the surface of the paintings is required, ablation with nanosecond (ns) pulses has been successfully implemented. However, despite significant advances, ns laser ablation presents limitations in dealing with the restoration of certain types of painted artworks, such as paintings with a very thin layer of varnish. On the other hand, femtosecond ( fs) laser processing seems to offer significant advantages compared to the case of longer pulse durations (Kruger et al. 2005, Hong et al. 2004). First, since multiphoton excitation is very efficient, the effective optical penetration depth is much reduced, the side-effects on the substrate being thus minimized (Kruger et al. 2005, Küper et al. 1987). In addition, thermal diffusion is limited, resulting in minimal heat ‘‘load’’ of the sublayers (Kruger et al. 2005).

93

LACONA8.indb 93

11/10/2010 5:36:41 PM

Since varnishes are complex systems, for a simplified approach we study polymers ((Poly(methyl methacrylate) PMMA) doped with aromatic compounds, the systems polymer/dopant being chemically similar to the painting materials used in artworks. A very important parameter for our study is polymer Molecular weight (Mw), which relates to the number of bonds that must be broken for material ejection to occur. In applications, Mw is a crucial parameter, since it can vary from layer to layer (different layers of varnishes can have different degrees of polymerization, depending on their exposure to various environmental conditions), or from one painting to another (the aging process affecting the Mw). 2

was found to result in reduced changes in the Raman spectra, multipulse irradiation was employed in order to obtain more conclusive results. In addition, a titling of the Raman spectra was observed for PhenI/PMMA systems (indicative of fluorescence). Therefore the Raman spectroscopy measurements presented here have been performed on neat PMMA (where no titling of the spectra has been observed), a higher precision of the measurements being thus ensured. 3

RESULTS AND DISCUSSIONS

Profilometric measurements show that, for ablation with 500 fs pulses at 248 nm, the single-pulse ablation thresholds and the etching rates are much reduced than those in the corresponding ns ablation, offering a better control of the etching/ cleaning process (Figure 1). Our finding is in agreement with previous studies (Küper et al. 1987), indicating that the ablation thresholds for ablation with 300 fs are much lower than those observed for ablation with 16 ns pulses. The differences between fs and ns UV laser ablation have been ascribed to the efficient operation of multiphoton processes, resulting in enhanced energy deposition within a smaller depth. Most importantly, for irradiation with UV fs pulses, the ablation threshold is found to increase, respectively the etching rates decrease with increasing Mw (Figure 2), as reported also in (Paun et al. 2009). Although similar features have been reported before in the ns ablation of polymers at 248 nm (Bounos et al. 2006, Rebollar et al. 2006), in the present case the dependencies are quantitatively different. Since the increase of the effective absorption with increasing fluence is similar for all

EXPERIMENTAL

Subpicosecond laser pulses (∼500 fs) at 248 nm were generated using a KrF laser system based on the principle of the Distributed Feedback Dye Laser (DFDL). The energy output was 10–30 mJ/ pulse with an average pulse-to-pulse fluctuation of 15%. The laser beam was focused perpendicularly onto the polymer by a convergent lens with 150 mm focal length, to an area of ∼0.2 mm2. All experiments were performed in air, repetition rate of 1 Hz being used. PMMAs of different molecular weights (Aldrich) were subjected to extensive purification. Iodophenanthrene (PhenI, Aldrich) was added to the polymer so that concentrations of 0.5% wt were obtained. The solutions of the purified polymers/dopants were cast on quartz pieces and left to dry for several hours in vacuum, resulting films of 20–30 μm thickness. The depth/profile of the ablated craters was measured with a profilometer (Perthometer®). Scanning Electron Microscopy (SEM) has been used to study the morphology of the irradiated areas. The chemical modifications induced by laser irradiation have been investigated by Raman spectroscopy. The Raman spectra have been recorded with a Raman system coupled with an optical microscope. The Raman system was equipped with a CCD camera and a laser operating at 473 nm (power level 15 mW) as an excitation source. The spectra have been recorded by collecting the scattered light from the surface in a backscattered geometry, with a resolution of 672 lines/mm and 60 s exposure time, the final spectra resulting from the accumulation of five individual ones. In order to avoid scattering of the excitation laser on any structures formed on the surface upon irradiation, UV laser irradiation has been effected at very low fluences. For assessing the chemical modifications upon irradiation, the samples have been irradiated with 500 Ds and 30 ns pulses at 248 nm. Since irradiation with few pulses

Ablation rate (µm/pulse)

30 ns pulse 15

500 fs pulse

10

5

0 0.0

0.5

1.0 1.5 Fluence (J/cm2)

2.0

Figure 1. Single-pulse ablation rate versus fluence for 0.5% wt PhenI/PMMA (199 kDa) for irradiation with a 500 fs, respectively 30 ns pulse at 248 nm. The ablation thresholds are indicated by vertical arrows.

94

LACONA8.indb 94

11/10/2010 5:36:41 PM

Ablation rate (µm/pulse)

2.5 2.5 kDa 2.0 199 kDa

1.5 1.0

996 kDa

0.5 0.0 0.0

0.4

0.8 1.2 1.6 2.0 Fluence (J/cm2)

2.4

Figure 3. (upper panel) SEM images of 0.5% wt PhenI doped PMMAs of the indicated Mws processed with a single fs pulse at 1.8 J/cm2; (lower panel) SEM images of 0.5% wt PhenI/PMMA (199 kDa) processed with a single fs pulse at the indicated fluences.

Figure 2. Single-pulse ( fs) ablation rate versus fluence for 0.5% wt PhenI/PMMA of the indicated Mws.

Mws examined (as shown by in-situ measurements of the laser intensity transmitted through the samples-data not shown here), the differences in the ablation thresholds and etching rates reflect differences in the efficiency of PMMA decomposition. Since the density of monomers is the same for all Mws, no particular dependence on Mw would be observed if side-group decomposition was the main factor determining material ejection. More likely, these dependencies indicate that material ejection involves main chain bond scission. Our results are supported by few previous studies (Baum et al. 2007) on gas chromatographic examination, combined with mass spectrometry, showing that, upon irradiation at 387 nm with ∼100 fs pulses, PMMA decomposition caused by scission of the main chain bonds results in the formation of the monomer (MMA). Turning to the investigation of the morphological modifications of the irradiated areas, SEM analysis indicates that upon irradiation with UV fs pulses, surface morphology differs markedly (being far superior) from the case of UV ns ablation. In the latter case, thermal decomposition of the polymers determins the growth of bubbles (Rebollar et al. 2005). In contrast, for single-pulse irradiation with fs pulses, for all Mws and at all fluences, the irradiated surfaces exhibit holes. Furthermore, we find that the size and number of the holes depend on polymer Mw, the holes being smaller in size and more numerous for the lower Mw PMMAs (Figure 3a, b, c). Interestingly, increasing fluence results only in the increase of the number density of the holes, while their size remains unchanged (Figure 3d, e, f). The mechanism responsible for the formation of the holes is not yet fully elucidated. One possibility is that the holes are formed by localized eruptions of gases as a result of polymer decomposition. This hypothesis is sustained by previous

studies, reporting MMA as the only gaseous product desorbing from PMMA upon irradiation with 100 fs pulses at 387 nm (Baum et al. 2007). Throughout this perspective, the fact that the size of the holes does not change (only their density increases) with increasing fluence is most likely due to the high efficiency of multiphoton absorption in the case of fs irradiation, resulting in much reduced effective optical penetration depth. Consequently, gas production is limited close to the surface, having a high probability of erupting. Alternatively, the holes might be formed by the influence of thermoelastic waves propagating within the polymer, caused by rapid temperature increase upon irradiation.Void formation (with a similar morphology and dependence on irradiation parameters as in our experimental results) has been modeled for simple molecular systems for irradiation with 15 ps pulses (Leveugle et al. 2004). Furthermore, similar effects have been modeled by molecular dynamics simulations, showing that upon irradiation of PMMA with 5 ps laser pulses, the temperature increase results in the propagation within the polymer of strong thermoelastic waves and determines its ejection (spallation) (Prasad et al. 2007). Thus, the increase of the holes size with increasing Mw is most probably due to the different viscoelasticity of different MWs. More precisely, higher Mw polymers exhibit higher viscoelesticity compared t the lower Mw polymers, and, therefore, their contraction (following spallation) into voids is slower. The observed dependences differ significantly from the ones reported for the ns irradiation (Rebollar et al. 2005, Paun et al. 2009). Upon fs irradiation, at the corresponding ablation thresholds we find the same number density of holes for all Mws examined (data not shown here). In contrast, in the case of ns ablation, the extent of bubble formation was more pronounced for

95

LACONA8.indb 95

11/10/2010 5:36:41 PM

the higher Mws (Rebollar et al. 2006), due to the higher temperatures attained in these polymers. These differences indicate that polymer decomposition upon fs irradiation is incompatible with the thermally initiated decomposition occurring in the UV ns ablation. A complete understanding of the mechanism of UV laser ablation of polymeric materials requires a systematic investigation of the chemical effects induced in the substrate upon irradiation. Until present, various vibrational spectroscopy techniques (FTIR, Raman spectroscopy) have been employed in order to assess the changes induced upon UV laser irradiation of polymers (as far as the chemical bonds are concerned). Most of the studies address the effects induced by laser irradiation with ns pulses (Rebollar et al. 2006, Beauvois et al. 1997). In contrast, very few studies have been focused on the irradiation with subpicosecond pulses. In order to overcome this lack of information, in the present study we report new results on the chemical modifications induced by UV fs/ns laser irradiation of PMMAs with different Mws using Raman spectroscopy. Examination of the Raman spectra of irradiated samples indicates the formation of a new band at 1640 cm−1 upon irradiation with both ns and fs pulses (Figure 4). Previous studies (F. Pallikari et al. 2001) have attributed this band to the stretching mode of the C = C bond (present only in the monomer), while the band at 1730 cm−1 was ascribed to the stretching mode of the C = O bond in the ester carbonyl group (present both in polymer and in monomer). Consequently, the ratio of the intensities of these two bands was used as a measure of the monomer content in the polymer. Similar considerations have been advanced by Bertoluzza and coworkers (Bertoluzza et al. 1987), suggesting that the band at 1640 cm−1 should be also attributed to the end-groups of

oligomer chains (more specifically to “unsaturation/ reactivity centers”). The formation of the C = C double bonds has been demonstrated by complementary techniques. FTIR measurements (Wochnowski et al. 2000) have indicated a peak at 1650 cm−1, that has been ascribed to a double bonding between two carbon atoms in the main chain. Both QMS and XPS spectra of PMMA irradiated at 248 nm at very low fluences have indicated the presence of methyl, methanol, carbon dioxide, demonstrating the complete side chain separation upon UV irradiation of PMMA. The complete side chain separation is followed by scission and formation of the double bonds observed in FTIR spectra. Furthermore, the formation of C = C double bonds upon irradiation of PMMA at 248 nm has also been demonstrated by picosecond ionization of the ejecta, indicating that the ejected fragments contain C = C bond, result which is confirmed by mass-spectrometric measurements. Conjugated structures containing C = C have been identified also in the LIF spectra (Andreou et al. 2002), by the emission band at 400–450 nm. Interestingly, our measurements indicate a dependence of the intensity of the Raman band at 1640 cm−1 on polymer Mw, with higher intensities for low Mw polymers. The fact that (independently of the decomposition mechanism) shorter chains are expected to decompose more efficiently may account for the stronger effect found for low Mw systems. Our result is supported by studies indicating that the intensity of the band at 1640 cm−1 increases with decreasing oligomer chain length (Bertoluzza et al. 1987). Most importantly, although the effects of UV laser irradiation on the Raman spectra are qualitatively similar for ns and fs pulses, a much weaker effect (smaller intensity of the band at 1640 cm−1) is observed following fs irradiation (Figure 5).

Figure 4. Raman spectra of PMMA (199 kDa) non-irradiated/irradiated at 50 mJ/cm2 with N = 150 pulses.

Figure 5. Raman spectra of PMMA (of the indicated Mws) irradiated at 50 mJ/cm2 with 500 fs pulses (N = 150 pulses).

96

LACONA8.indb 96

11/10/2010 5:36:41 PM

Bertoluzza, A., Garcia-Ramos, J.V., Fagnano, C., Monti, P., Semerano, G., Garcia-Ramos, J.V., Caramazzaf, R. & Cellini, M. 1987. Raman spectra of intraocular lenses before and after implantation in relation to their biocompatibility. Journal of Raman Spectroscopy 18: 151–152. Bounos, G., Selimis, A., Georgiou, S., Rebollar, E., Castillejo, M. & Bityurin, N. 2006. Dependence of ultraviolet nanosecond laser polymer ablation on polymer molecular weight: Polymethyl methacrylate at 248 nm. Journal of Applied Physics 100: 114323. Georgiou, S., Zafiropulos, V., Anglos, D., Balas, C., Tornari, V. & Fotakis, C. 1998. Excimer laser restoration of painted artworks: Procedures, mechanisms and effects. Applied Surface Science 738: 127–129. Heitz, J., Arenholz, E., Bäuerle, D., Sauerbrey, R. & Phillips, H.M. 1994. Femtosecond Excimer Laser Induced Structure Formation on Polymers. Applied Physics A 59: 289–293. Hong, M.H., Luk’yanchuk, B., Huang, S.M., Ong, T.S., Van, L.H. & Chong, T.C. 2004. Femtosecond laser application for high capacity optical data storage. Applied Physics A 79: 791–794(4). Kruger, J., Martin, S., Madebach, H., Urech, L., Lippert, T., Wokaun, A. & Kautek, W. 2005. Femtoand nanosecond laser treatment of doped polymethylmethacrylate . Applied Surface Science 247: 406–411. Küper, S. & Stuke, M. 1987. Femtosecond UV excimer laser ablation. Applied Physics B 44: 199–204. Leveugle, E., Ivanov, D.S. & Zhigilei, L.V. 2004. Photomechanical spallation of molecular and metal targets: molecular dynamics study. Applied Physics A 79: 1643–1655. Pallikari, F., Chondrokoukis, G., Rebelakis, M. & Kotsalas, Y. 2001. Raman spectroscopy: A technique for estimating extent of polymerization in PMMA. Mat Res Innovat 4: 89–92. Păun, I.A., Selimis, A., Bounos, G., Kecskeméti, G. & Georgiou, S. 2009. Nanosecond and Femtosecond UV Laser Ablation of Polymers: Influence of Molecular Weight. Applied Surface Science doi: 10.1016/ j.apsusc.04.106. Prasad, M., Conforti, P.F. & Garrison, B.J. 2007. Journal of Chemical Physics 127 084705: 1–13. Preuss, S., Spath, M., Zhang, Y. & Stuke, M. 1993. Time resolved dynamics of subpicosecond laser ablation. Applied Physics Letters 62: 3049–3051. Rebollar, E., Bounos, G., Oujja, M., Domingo, C., Georgiou, S. & Castillejo, M. 2005. Influence of polymer molecular weight on the UV ablation of doped poly(methyl methacrylate). Applied Surface Science 248: 254–258. Rebollar, E., Bounos, G., Oujja, M., Georgiou, S. & Castillejo, M. 2006. Influence of polymer molecular weight on the UV ablation of doped poly(methyl methacrylate). Jurnal of Physical Chemistry B 110: 16452. Wochnowski, C., Metev, S. & Sepold, G. 2000. UVLaser-Assisted Modification of the optical properties of Polymethylmethacrylate. Applied Surface Science 154–155: 706–711. Zafiropulos, V. & Fotakis, C. 1998. Laser Cleaning in Conservation: An Introduction. Oxford: M. Cooper (Ed.) Butterworth Heinemann.

This indicates that the polymer decomposition in fs irradiation regime differs from the one in the ns case. More specifically, while in the ns regime a thermal mechanism was shown to dominate, upon irradiation with fs pulses the laser induced chemical modifications are significantly reduced, due to the high efficiency of multiphoton processes. 3

CONCLUSIONS

In all, fs ablation ensures higher etching resolution, superior surface morphology and considerably reduced chemical modifications in the remaining substrate, therefore presenting high potential for the restoration of paintings that are not amenable to restoration with nanosecond lasers. Furthermore, a strong influence of the polymer Mw was found on the morphology, on the etching rates, as well as on the extent of chemical modifications, demonstrating that ablation with UV fs laser pulses involves scission of the main chain bonds. Besides the implications concerning the ablation mechanism, elucidation of the influence of the Mw on polymer UV laser processing allows the optimization of the irradiation protocols according to the characteristics of the sample. ACKNOWLEDGEMENTS The work was supported by the Ultraviolet Laser Facility operating under the Improving Human Potential (IHP)-Access to Research Infrastructures program (RII3-CT-2003-506350), by PENED 01ED419 and by PENED 03ED351 administered by the Greek Ministry of Development. The authors wish to thank O. Kokkinaki for useful discussions and for corrections of this manuscript. I-A.P. acknowledges the support of the EU Marie Curie Early Stage Training Projects ATHENA (MEST-CT-2004-504067) and ATLAS (MEST-CT-2004-008048). REFERENCES Andreou, E., Athanassiou, A., Fragouli, D., Anglos, D. & Georgiou, S. 2002. Laser and Material parameter dependence of the chemical modifications in the UV laser processing of model polymeric solids. Laser Chemitry 20(1): 1–21. Baum, A., Scully, P.J., Basanta, M., Paul Thomas, C.L., Fielden, P.R., Goddard, N.J., Perrie, W. & Chalker, P.R. 2007. Photochemistry of refractive index structures in poly(methyl methacrylate) by femtosecond laser irradiation. Optics Letters 32: 190–192. Beauvois, S., Renaut, D., Lazzaroni, R., Laude, L.D. & Bredas, J.G.L. 1997. Physico-chemical characterization of the effect of excimer laser irradiation on PMMA thin films. Applied Surface Science 109–110: 218–221.

97

LACONA8.indb 97

11/10/2010 5:36:42 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Monitoring the laser cleaning process of ornamental granites by means of digital image analysis J. Lamas, A.J. López & A. Ramil Centro de Investigacións Tecnolóxicas, Universidade da Coruña, Ferrol, Spain

B. Prieto Dpto. Edafoloxía e Química Agrícola, Universidade de Santiago de Compostela, Santiago de Compostela, Spain

T. Rivas Dpto. Enxeñaría dos Recursos Naturais e Mediambiente, Universidade de Vigo, Vigo, Spain

ABSTRACT: A monitoring technique based on digital image analysis has been developed for laser cleaning of granites. In brief, the monitoring system consists on a digital photo camera to capture the image of the granite surface under proper lighting before and after laser irradiation. The analysis of the images, based on artificial neural networks, provides accurate L*a*b* color coordinates of the surface which were used as indicators for monitoring. Different scenarios comprising the removal of biological crust, spray paint (graffiti) or the irradiation of fresh samples were analyzed and results show that color coordinates obtained by image analysis provide not only a clear indication of the surface cleanliness but also surface damage. 1

INTRODUCTION

distance between two different colors corresponds approximately to the color difference perceived by the human eye. Artificial neural networks can be used to obtain accurate device-independent L*a*b* color units from device dependent RGB values captured by a digital color camera (Leon et al. 2006). Different scenarios have been analyzed which show the capability of L*a*b* color coordinates obtained by image analysis to monitor the process: laser removal of biological black crusts and spray paint (graffiti) and, moreover surface damage in the form of color changes in fresh granite samples under laser irradiation.

Laser cleaning is a well established technique in the field of Cultural Heritage, and different monitoring approaches based on spectroscopy (GobernadoMitre et al. 1997) and acoustic or chromatic techniques (Lee et al. 2000; Strlic et al. 2005) have been developed to asses an effective cleaning without damaging the substrate. Some authors have shown that color is an essential and highly vulnerable feature in the process of laser cleaning of stones (Grossi et al. 2007, Pouli et al. 2008).Usually the color variations on stones associated with different operative laser fluences are measured using a colorimeter which sample points a few mm in diameter only. They are thus largely irrelevant for monitoring color variations in textured materials like granites whose poly-mineral composition results in high chromatic heterogeneity (Lebrun et al. 2004). The aim of this work is to propose a methodology for monitoring the cleaning of granite surfaces and to quantify color changes caused by laser irradiation using digital image analysis. By means of a digital camera it is possible to register the RGB values of any pixel of the image of the object. RGB is a non-absolute color space; however CIEL*a*b* space is perceptually uniform, i.e., the Euclidean

2

MATERIALS AND METHODS

The granites used in this study are denominated Vilachán and Rosa Porriño. Vilachán is a finegrained and brownish-yellow colored stone which is extensively used as ornamental granite in areas of Galicia (NW Spain). Rosa Porriño, one of the most marketed ornamental granites in Spain; is a medium to coarse grained rock and presents a characteristic pink colour. Laser treatments were carried out by the third harmonic, 355 nm, of both Q-switched Nd:YAG

99

LACONA8.indb 99

11/10/2010 5:36:42 PM

and Nd:YVO4 laser sources. Nd:YAG is a Quantel, model Brilliant B, with frequency f = 10 Hz, pulse duration 6 ns and maximum pulse energy varying between 10–60 mJ. Nd:YVO4 is a Coherent AVIA Ultra 355–2000 with pulse duration 25 ns and a pulse repetition rate that can be selected from single-shot to 100 kHz with a energy per pulse around 0.1 mJ. Samples were mounted onto a 3D translation stage and submitted to different irradiation conditions depending on the purpose of the laser treatment. Figure 1 shows a schematic diagram of the experimental set up. After the laser treatment, granite samples were examined by optical microscopy to asses the cleaning efficiency and to detect morphological changes in the stone surface. Color measurements were performed on every sample before and after laser irradiation by using a digital camera Nikon D100 with exposure times ranging from 30 s to 1/4000 s and an objective Nikon micro 60/2.8. The CCD of the camera consists of 6.1 × 106 pixels and presents an active area of 23.7 × 15.6 mm2. The images were stored in non-compressed file (TIF format) to avoid loss of image quality. The standard light source D65 (ISO 11664-2: 2007) which mimic variations of daylight and with a color temperature of 6500 K was used. The position of the camera, sample, and light sources was arranged in order to capture the diffuse reflection responsible for the color, which occurs at 45° from the incident light, and to ensure a uniform distribution of light intensity over the sample (Ramil et al. 2008a). A standard colored chart GretagMacbeth ColorChecker was used to calibrate, verify the experimental settings prior to actual measurements and also for training the neural network. A colorimeter Color-Eye-XTH with a 10 mm diameter measuring head was used to obtain L*a*b* color values for the training of the neural network. In this work, firstly we have considered the removal of biological black crust and spray paint (graffiti) in samples of Vilachán, and, secondly, the loss of coloration of Rosa Porriño under laser irradiation; both scenarios are discussed below.

y Laser

z

x

Figure 1. Experimental set up for the laser cleaning processes.

3

DATA PROCESSING

For the transformation of RGB values into L*a*b*, a feed-forward back-propagation neural network with three layers was designed (Haykin, 1999). Given that the input vector consisted of RGB values of each pixel of the image and the outputs were the L*a*b* variables estimated from the network; both the input and output layers consisted of three neurons, each one corresponding to one color coordinate. The number of neurons in hidden layer was optimized to 3 and the training was stopped at 5 × 106 cycles. Details about the process of optimization and training can be found in a previous work (Ramil et al. 2008b). In the present study, L*a*b* coordinates of 32 color charts of the GretagMacbeth ColorChecker were measured using the colorimeter. Additionally, a RGB digital image was taken of each chart. Each chart was divided into 25 regions and the R, G and B color values of the corresponding regions were measured using a Matlab® program which computes the mean values for each color value in each region (Westland et al. 2004). Thus, 800 RGB measurements were obtained. The set of RGB values of the color chart were divided into three groups; one for the calibration or training of the network (50%), other for optimization (25%) and the rest for validation purposes (25%). The complete procedure was designed and performed using the Matlab® Neural Network Tool-box (Demuth et al. 1992). 4

RESULTS

4.1 Laser removal of black patinas We present here the removal of black layers from the granite surface; i.e. biological black crust and black spray paint. In both cases the soiling layer is darker than the mean color of the substrate. Firstly we considered natural exposed samples of granite Vilachán, covered with a layer of biological black crust (thickness around 10–30 μm). The cleaning was performed by means of Nd: YVO4 laser operating at the wavelength of 355 nm. Figure 2 depicts one of the cleaning tests performed. The black contour is the biological crust; dark spots in the cleaned area correspond to biotite (black mica) grains. The analysis of digital images of the cleaned test areas to obtain L*a*b*, allowed us to control the process of crust removal. Figure 3 shows the values of lightness (L* coordinate) as a function of the laser fluence in the range 0.1–2.4 J/cm2. As can be seen from the plot, L* rises steeply as the fluence increases from 0.1 to 0.3 J/cm2; then remains steady in the range 0.3 J/cm2 to 1.2 J/cm2 and, finally, it shows a new increment.

100

LACONA8.indb 100

11/10/2010 5:36:42 PM

Figure 2. Laser cleaning test in a sample of granite Vilachán covered with a biological black crust. Figure 4. Detail of the image obtained by optical microscopy of the textural changes observed in granite Vilachán at the fluence of 1.9 J/cm2. Scratches in the feldspar grains and erosion of quartz crystals are appreciable.

70 60 50 L*

65

40 60

30 55 L*

20 -1

10 F/[J/cm 2]

45 40

Figure 3. Removal of biological black crusts in granite Vilachán. L* coordinate as a function of the laser fluence.

35

Inspection of laser treated areas by optical microscopy revealed that fluences above 0.5 J/cm2, which correspond to the plateu of the plot, give rise a clean surface without any biological structure. The increase of lightness at the highest fluences was associated to textural changes observed in the stone surface (scratches in the feldspar grains and erosion of quartz crystals) (Figure 4). The removal of black paint layers can be also monitored by L*. Figure 5 depicts the lightness L* as a function of the overlapping coefficient K. This coefficient is defined as the ratio between the beam diameter, d, and the displacement of the sample between two consecutive shots, v/f (being v the speed of scan and f the laser repetition frequency), then. K=

d f v

50

0

10

(1)

0

2

4 K = d ⋅ f/v

6

8 4

x 10

Figure 5. L* coordinate as a function of the overlapping coefficient K in the removal of black paint from the granite Vilachán surface.

In order to address an efficient cleaning of large areas of the stone surface, it is necessary to determine the optimum value of K. In this case, after a rapid increase, the lightness L* starts to saturate at values of the overlapping coefficient K ≈ 4 × 104, which indicates the optimum value of K to address the complete removal of the black paint. 4.2 Laser removal of colored graffiti In contrast with the cleaning of black paint, the removal of different paint colors in granite Vilachán requires not only the knowledge of L* but also the chromatic parameters a* and b*.

101

LACONA8.indb 101

11/10/2010 5:36:42 PM

This is because the lightness of the dark minerals of the granite surface and the lightness of the paint are now comparable. To attain an even coverage of the surface that ensures the complete removal of the paint, different laser trajectories were programmed. Figure 6 corresponds to the removal of a red paint layer. The values of color coordinates a* and b* were obtained by scanning the laser beam along a series of parallel lines (a), and along orthogonal linear paths, (b). Case (c) corresponds to non-painted granite. In the former case the variability of a* and b* in fresh samples can be appreciated. Ellipses in the plot correspond to one standard deviation (±σ) of the distribution of chromatic parameters; i.e., 68% of the values of a* and b*. As the cleaning improves, a* and b* decrease, and the ellipses approach the case (c) which corresponds to a cleaned surface. Figure 7 depicts the removal of a red paint layer in granite by means of scanning the laser along orthogonal linear paths. The effectiveness of the cleaning can be appreciated. 4.3

Figure 7. Image of the removal of red paint from granite surface. The red layer (left side) was completely removed (right side) by scanning the laser along orthogonal paths. Dark spots in the cleaned area correspond to biotite grains. 16 14

Color changes in ornamental granite Rosa Porriño

12 10

1.03 J/cm2

8

0.76 J/cm2

b*

Color changes in Rosa Porriño under UV: Nd:YAG laser irradiation were previously reported (Ramil et al. 2008a). These changes, appreciable under naked eye examination, were manifested as a loss of pink coloration which affected the chromatic coordinates but not lightness (no discernible trend was appreciated in L*). Figure 8 depicts the changes in chromatic coordinates a* and b*, obtained by means of digital

0.53 J/cm2

6

0.35 J/cm2 0.18 J/cm2

4 2 -2

0.00 J/cm2 0

2

4

6

8

a*

14

Figure 8. laser.

12

b*

10

image system, as a function of the laser fluence. First, it must be noticed the high dispersion in values of a* and b* in the case of fresh granite, highlighting the problem of colour measurement in this stone types by means of conventional colorimetric techniques. Secondly, the plot depicts the loss of coloration as the laser fluence increases. The color of Rosa Porriño granite becomes more uniform (reduction of the ellipse size), a* and b* decrease and the mean colour of the stone tends to gray.

8 6 (a) (b) (c)

4 2

0

2

4

6

8

Color changes in Rosa Porriño granite under

10

a*

Figure 6. Laser removal of red spray paint from granite Vilachán. Chromatic parameters a* and b* obtained scanning the laser beam along a series of parallel lines (a), along orthogonal linear paths, (b). Case (c) corresponds to non-painted granite.

5

CONCLUSIONS

A monitoring system based on the analysis of digital images has been developed for the laser cleaning

102

LACONA8.indb 102

11/10/2010 5:36:43 PM

of granites. Accurate L*a*b* color coordinates of the surface obtained through neural network logic were used as indicators for monitoring. The polymineral composition of granites results in high chromatic heterogeneity which makes irrelevant the use of colorimeters for monitoring color variations. It was found that the color coordinates obtained through the analysis of digital images were able to track the change of surface condition under laser irradiation. In the case of patinas darker than the substrate (biological black crust, or black paint) the lightness, L* coordinate, can be used to monitoring the laser process; in other cases, knowledge of chromatic coordinates a* and b* is required to attain an effective cleaning and to monitor the damage. This monitoring system is fast, cost effective and can be also applied to the laser cleaning of different materials. ACKNOWLEDGMENTS Work funded by Galician Government Research Project PGIDIT06CCP00901CT. REFERENCES Demuth, H., M. Beale, and M. Hagan (1992). Neural Network Toolbox User’s Guide (2007 ed.). The Mathworks Inc. Gobernado-Mitre, I., A.C. Prieto, V. Zafiropulos, Y. Spetsidou and C. Fotakis (1997). On-line monitoring of laser cleaning of limestone by laser-induced breakdown spectroscopy and laser-induced fluorescence Applied Spectroscopy 51, 1125–1129. Grossi, C.M., F.J. Alonso, R.M. Esbert, and A. Rojo (2007). Effect of laser cleaning on granite color Color Research and Application 32(2), 152–159.

Haykin, S. (1999). Neural Networks: A comprehensive foundation. New Jersey: Prentice Hall. Lebrun, V., C. Toussaint and E. Pirard (2004). Monitoring color alteration of ornamental flagstones using digital image analysis. In R. Prikryl (ed.) Dimension stone 2004: 139–145. London: Taylor and Francis. Lee, J.M. and K.G. Watkins (2000). In-process monitoring techniques for laser cleaning Optics and Lasers in Engineering 34 (4–6) 429–442. Leon, K., D. Mery, F. Pedreschi, and J. Leon (2006). Color measurement in L*a*b* units from RGB digital images. Food Research International 39(10), 1084–1091. Pouli, P., C. Fotakis, B. Hermosín, C. Saiz-Jimenez, C. Domingo, M. Oujja and M. Castillejo (2008). The laser- induced discoloration of stonework; a comparative study on its origins and remedies Spectrochimica Acta Part A 71 932–945. Ramil, A., A.J. López, M.P. Mateo, C. Álvarez, and A. Yáñez (2008a). Colour changes in Galician granitic stones induced by UV Nd:YAG laser irradiation. In: M. Castillejo et al. (eds.), LACONA VII Proceedings: 199–202. New York: CRC Press-Taylor & Francis Group. Ramil, A., A.J. López and A.Yañez (2008b). Application of artificial neural networks for the rapid classification of archaeological ceramics by means of Laser Induced Breakdown Spectroscopy (LIBS) Applied Physics A 92 197–202. Strlic, M., V.S. Selih, J. Kolar, D. Kocar, B. Pihlar, R. Ostrowski, J. Marczak, M. Strzelec, M. Marincek, T. Vuorinen and L.S. Johansson (2005). Optimization and on-line acoustic monitoring of laser cleaning of soiled paper Applied Physics A 81 (5) 943–951. Westland, S.W. and C. Ripamonti (2004). Computational colour science using Matlab. England: John Wiley & Sons.

103

LACONA8.indb 103

11/10/2010 5:36:46 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Optimization of laser cleaning parameters for the removal of biological black crusts in granites A.J. López, J. Lamas, A. Ramil & A. Yáñez Centro de Investigacións Tecnolóxicas, Universidade da Coruña, Ferrol, Spain

T. Rivas & J. Taboada Dpto. Enxeñaría dos Recursos Naturais e Medioambiente, Universidade de Vigo, Vigo, Spain

ABSTRACT: Laser cleaning of stones is a well established technique in the field of cultural heritage, however there are few studies concerning its application to granites. Galicia (NW Spain) has a rich heritage of buildings and monuments constructed from locally obtained granite. In humid climates granite is almost permanently damp which causes biological colonization and blackening of exterior surfaces. This work is focused on the removal of biological black crust from granite by means of Nd:YVO4 laser at the wavelength of 355 nm. Analysis of the conditions for efficient removal without causing damage in stone surface were performed by means of a number of analytical techniques (optical microscopy, optical profilometry, SEM-EDX) which allowed us to establish safe conditions of irradiation. 1

INTRODUCTION

Galicia (NW Spain), whose geological substrate is mainly of granitic rocks, has a rich heritage of buildings and monuments constructed from locally obtained granite. Because of the humid climate in Galicia, the granite is almost permanently damp which favors biological colonization and blackening of exterior surfaces. A complete description of the black crust from petrographical, chemical, mineralogical and biological point of views can be found elsewhere (Prieto et al. 2007). In addition to aesthetic considerations, it is well known that biological factors contribute to the weathering of the rocky substrate through both physical and chemical interactions (Warscheid et al. 2000). Although laser cleaning of stonework has become a well established technique; much published work concerns limestone and marbles, and to a lesser extent, silicate rocks such as sandstone, these being the most widely used as building materials for sculptures and monuments in the countries that have pioneered studies of conservation of historic monuments (Fotakis et al. 2007). To the best of our knowledge, there are few works concerning the laser cleaning of granites, (Wakefield et al. 1997, Grossi et al. 2007, Pan et al. 2009) and none of them dealing with the removal of biological crusts.

This work is part of a multidisciplinary project involving scientist and conservators with the aim of analyzing the state of deterioration of The Ruins of Santo Domingo Church (Pontevedra, Galicia) and to propose conservation procedures. The Ruins consist of five apses of the fourteenth and fifteenth centuries, exceptional in Galician gothic architecture, with stone tombs of the nobility, capitals of medieval sculpture and coats of arms. The Ruins are nowadays part of the Museo de Pontevedra. Many elements of this architectural group, especially those situated in the Northern side present extensive biological colonization and blackening as can be appreciated in Figure 1. The objective of the present work is to analyze the laser removal of biological black crust from the granite surface. Other authors (Marakis et al. 2003) have found that the 3rd harmonic of a Nd:YAG was very successful in removing biological encrustation on marble, owing to the high absorption of UV light by biological molecules; however, the polymineral composition of granite makes it impossible to extrapolate results obtained with other rocks. In this work we have used a Nd:YVO4 laser at the wavelength of 355 nm. In order to optimize the cleaning parameters we have focused our analysis on the possible morphological/textural changes caused by the laser on the stone surface evaluated by means of different surface analytical techniques.

105

LACONA8.indb 105

11/10/2010 5:36:46 PM

Figure 1. Coat of arms situated in the Northern side of the Ruins of Santo Domingo Church (Pontevedra, Galicia) showing intense biological colonization and blackening.

2

EXPERIMENTAL

Experiments were carried out on a granite denominated Vilachán, a fine-grained and brownishyellow colored stone which is extensively used as ornamental granite in the province of Pontevedra. This granite is, from a mineralogical and textural point of view, similar to the fabric of the Ruins. The essential minerals are quartz (47%), K-feldspar (10%), plagioclase (15%), muscovite (18%) and biotite (7%). Natural exposed rock fragments showing intense surface blackening were selected for this study. Visualization by SEM shows that the crust presents characteristics similar to biological deposits in samples of granite rocks in non-urban environments (Prieto et al. 2007). The crust is observed as a continuous patina of variable thickness (between 10 and 35 μm) formed by spherical particles which correspond to collapsed cells of algae and filamentous structures which can be attributed to fungal hyphae (Fig. 2a). Furthermore, minerals showing exfoliation planes (like feldspars or mica) oriented perpendicularly to the surface favor the crust developing to a greater depth in the rock (Fig. 2b). The laser used in this work was a Nd:YVO4 (Coherent AVIA Ultra 355–2000) at the wavelength of 355 nm and pulse duration 25 ns. The intensity profile at the laser output was near-Gaussian (M2 < 1.3) and the beam diameter at 1/e2 intensity level was about 2.2 mm. The pulse repetition rate can be selected from single-shot to 100 kHz with a energy per pulse around 0.1 mJ. The samples were set on 3D translation stage Newport ILS-CC.

Figure 2. SEM characterization of the biological black crust: (a) Image of colonized surface obtained at SE mode. (b) Image of a transversal cut obtained at BE mode.

The focused beam was aimed approximately normal to the sample surface. The sample surface was precisely positioned at the beam waist of the focused beam using the z-direction stage. In order to select the adequate parameters of cleaning, a set of preliminary test were performed both in soiled granite and granite without crust as control. From these tests we have observed that the black crust highly increases the absorption of the laser energy, thus, at fluences considered safe in the case of granite without crust appreciable damage is caused in the granite with crust. Since our objective was the removal of the biological black crust, we have focused our study on samples covered with crust. Thus, after trying different working parameters i.e., repetition frequency, relative position between sample and laser, or speed of scan, we have selected a fixed frequency of 10 kHz, and fluences in the range 0.1 to 2.5 J/cm2 as the most adequate for our purpose; while the speed of scan and the separation between adjacent lines were kept constant, 20 mm/s and 75 μm, respectively. Analysis of the stone surface after laser irradiation was carried out by means of various techniques: Optical microscopy (Nikon Eclipse L150) to obtain the first evaluation of the cleaning

106

LACONA8.indb 106

11/10/2010 5:36:46 PM

process; optical profilometry (Wyko-NT 1100) using white light Vertical Scanning Interferometry (VSI) mode was used for the topographical characterization of the surface; and SEM/EDX, (Phillips XL30) both in secondary (SE) and backscattered (BE) electron modes gave morphological/chemical information about the black crust and the effects of the laser treatment on the granite rock-forming minerals. Finally, X-ray photoelectron spectroscopy (XPS, ESCALAB 250iXL, VG Scientific) provided chemical analysis of the outermost layers of the stone material. 3

RESULTS AND DISCUSSION

An image of the granite sample showing a set of cleaning tests performed at fluences in the range 0.1–2.4 J/cm2 is shown in Figure 3. It can be visually appreciated the improvement of the cleaning quality at increasing fluence. Evaluation by optical microscopy revealed a cleaned surface, without any rest of biological structure at fluence values ≥0.5 J/cm2, nevertheless certain textural changes in the surface were clearly visible at the highest fluences (1.9 J/cm2–2.4 J/cm2). These textural changes appear as a pattern of orthogonal scratches, mainly in the surface of feldspars grains; in addition with a clear erosion of quartz crystals. At this point, it seems clear that the optimization of the cleaning process requires further knowledge of the critical region comprised between 0.7 J/cm2 and 1.9 J/cm2, where, under these irradiation conditions, the damage threshold seems to be located. Topographical analysis of the surface was based on the probability distribution of surface heights and its characteristic parameters Ra, Rq, Rsk and Rku, obtained from 3-D images given by optical profilometry. Mathematical definition of these parameters can be seen elsewhere (Gadelmawla et al. 2002). In brief, Ra represents the average absolute deviation height and Rq the standard deviation of the distribution; the skewness, Rsk, measures

the symmetry of the profile about the mean line (a symmetrical height distribution has zero skewness); and kurtosis, Rku, describes the sharpness of the distribution. A surface with a random probability distribution has Rsk = 0 and Rku = 3. The values obtained both in the soiled surface and laser treated areas are included in Table 1. The former case is close to a random distribution. The removal of the biological crust at 0.7 J/cm2 is reflected in a major presence of negative values in the distribution of heights (Rsk = −1) due to the emptying of the voids in the stone surface; in addition with a higher roughness, Rq = 6.3 μm. Finally, laser treated areas at fluence 1.9 J/cm2 present the highest roughness (Rq = 9.6 μm); a height distribution with lower kurtosis (Rku = 2.9) and a skewness nearby zero (Rsk = 0.6), which indicates a clear reduction of the peak heights reflecting the erosion of the stone surface caused by the laser. To visualize distinctive features of the laser tread, line profiles obtained from 3-D topographical images of the interface soiled-cleaned are depicted in Figure 4 and Figure 5 where z-coordinate represents the height. At 0.7 J/cm2 (Fig. 4) the ablation depth in the crust is around 20 μm, however, at the fluence of 1.9 J/cm2 (Fig. 5), the line profile shows the width and depth of the scratches caused by the laser on the granite substrate (width ≈ 40 μm, depth ≈ 5 μm). Owing to the poly-mineral composition of granite, the differential effect of the laser irradiation on the rock-forming minerals has been examined by scanning electron microscopy. Figure 6 and Figure 7 show SEM micrographs at Backscattered Electron mode (BE) which allowed visual distinction of the minerals quartz, feldspar and biotite. Both images summarize the laser effects at energy densities 0.7 and 1.9 J/cm2, respectively. There is no evidence of scratching in the surface exposed to the lowest fluence (Fig. 6), however, the most striking feature is the melting of biotite grains. This effect was also seen by other authors using a 1064 nm laser source (Grossi et al. 2007). The scratches caused by the laser at the highest fluence (Fig. 7) are manifested to a greater or lesser extent depending on the mineral, so feldspar seems to be the hardest one and depicts the above mentioned orthogonal pattern, laser treads are also appreciated on biotite grains, while quartz Table 1. Characteristic parameters of the probability distribution of surface heights.

Figure 3. Granite sample showing a set of cleaning tests performed at fluences in the range 0.1–2.4 J/cm2.

Sample

Ra (μm)

Rq(μm)

Rsk

Rku

Soiled 0.7 J/cm2 1.9 J/cm2

2.51 4.97 7.99

3.01 6.3 9.64

−0.08 −1.06 −0.6

2.41 4.54 2.93

107

LACONA8.indb 107

11/10/2010 5:36:46 PM

Figure 6. SEM image of the granite surface of laser treated area at 0.7 J/cm2. Q, quartz; F, feldspar; B, biotite.

Figure 4. Representative 3D-image of the interface soiled-cleaned and corresponding line profiles at fluence 0.7 J/cm2. The ablation depth is around 20 μm.

Figure 7. SEM image of the granite surface of laser treated area at 1.9 J/cm2. Q, quartz; F, feldspar; B, biotite.

Figure 5. Representative 3D-image of the interface soiled-cleaned and corresponding line profiles at fluence 1.9 J/cm2. The scratches caused by the laser on the granite substrate can be appreciated (width ≈ 40 μm, depth ≈ 5 μm).

shows thermally related brittle fracture. At higher magnification an intense pitting of feldspars grains is also appreciable (Fig. 8), and characteristic cleavage stepped fracture of quartz in addition with ovoid structures similar to drops which evidences the biotite melting are shown in Figure 9. Similar structures were observed by other authors in granites submitted to high temperatures during the process of flamed finish (Rojo et al. 2003).

Figure 8. Detail of textural changes caused by the laser at 1.9 J/cm2 in feldspar grains. An intense pitting can be appreciated.

SEM examination also confirmed the absence of organic structures in laser treated areas, however chemical analysis by EDX detected weak signals of Carbon, probably masked by the strong signals of characteristic elements of the stone (Al, K, Si, Fe …). XPS spectra surveys were

108

LACONA8.indb 108

11/10/2010 5:36:46 PM

Figure 9. melting.

Detail of brittle fracture of quartz and biotite

Table 2. C/O atomic ratio (%) obtained in granite samples by means of XPS spectra (survey scan). Sample

C/O ratio (%)

Non-cleaned area Cleaned at fluence 0.7 J/cm2 Cleaned at fluence 1.9 J/cm2 Granite without crust

4.55 ± 1.04 1.43 ± 0.29 0.68 ± 0.09 0.97 ± 1.17

ACKNOWLEDGMENTS This work was funded by Galician Government Research Projects: PGIDIT06CCP30401PR and PGIDIT06CCP00901CT. A.J. López and T. Rivas contributions were also financed by Cátedra Filgueira Valverde (Universidade de Vigo) through project “Estudio de los factores de deterioro y propuestas de conservación de las Ruinas de Santo Domingo”.

performed in laser treated areas. A non-cleaned area (i.e. black crust) and natural granite sample without crust were also examined to be used as reference. Table 2 presents the C/O atomic ratio which was obtained as the average value of four measurements in each surface. As it can be appreciated, laser cleaning at the fluence of 0.7 J/cm2; led to a clear reduction of the C/O ratio, compatible with the value obtained in samples of natural granite without crust. This result corroborates the efficiency of the cleaning process observed by SEM. As expected, given that the highest fluence 1.9 J/cm2 causes the removal of layers of the granite substrate, the C/O ratio is even lower than in surface without crust. 4

threshold was located around 1.5 J/cm2; above this value morphological/textural changes in the stone surface were appreciated and could be evaluated by means of optical profilometry. Owing to the poly-mineral composition of granite, differential behaviour of the constituent minerals under laser was assessed by SEM: biotite was the most affected mineral in the sense that it reached melting even at fluences that were considered safe (≤1.5 J/cm2). Though biotite content is low, and thus the damage restricted, further analysis is under way to adjust the irradiation parameters in order to diminish heat accumulation at the stone surface which causes such undesirable thermal effect.

CONCLUSIONS

In the present work we have analyzed the removal of biological black crust in granite Vilachán by means of a Nd:YVO4 laser operating at the wavelength of 355 nm. The crust is appreciated as a black patina of variable thickness which tends to penetrate into the pores of the rock surface. At fixed frequency of 10 kHz, fluences in the range of 0.1–2.5 J/cm2 were tested and results led to values ≥0.5 J/cm2 as appropriate to ensure the complete removal of the biological crust. Further evaluation of the stone surface by different analytical techniques allowed us to ensure that the damage

REFERENCES Fotakis, C., Anglos, D., Zafiropulos, V., Georgiou, S. & Tornari, V. 2007. Lasers in the Preservation of Cultural Heritage. Principles and Applications. New York, London: Taylor & Francis. Gadelmawla, E.S., Koura, M.M., Maksoud, T.M.A., Elewa, I.M. & Soliman, H.H. 2002. Roughness parameters, Journal of Materials Processing Technology. 123: 133–145. Grossi, C.M., Alonso, F.J.¸ Esbert, R.M. & Rojo, A. 2007. Effect of laser cleaning on granite color, Color Research and Application 32: 152–159. Rojo, A., Alonso, R. & Esbert, R.M. 2003. Hydric propertiesof some Iberian ornamental granites with different superficial finishes: a petrophysical interpretation, Materiales de Construccion 53: 61–72. Marakis, G., Pouli, P., Zafiropulos, V. & MaravelakiKalaitzaki, P. 2003. Comparative study on the application of the 1st and the 3rd harmonic of a Q-switched Nd:YAG laser system to clean black encrustation on marble, Journal of Cultural Heritage 4: 83 s–91 s. Pan, A., Chiussi, S., Serra, J., González, P. & León, B, 2009. Excimer laser removal of beeswax from galician granite monuments, Journal of Cultural Heritage 10: 48–52. Prieto, B., Aira, N. & Silva, B. 2007. Comparative study of dark patinas on outcrops and buildings, Science of Total Environment 381: 280–289. Wakefield, R.D., Brechet, E. & McStay, D. (1997). The effect of laser cleaning on Scottish granite. Lasers as Tools for Manufacturing 2 2993, 246–251. Warscheid, Th. & Braamsb, J. 2000. Biodeterioration of stone: A review, International Biodeterioration & Biodegradation 46: 343–368.

109

LACONA8.indb 109

11/10/2010 5:36:47 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Bronze putti from Wilanów Palace garden façade—conservation studies and tests of laser cleaning H. Garbacz, E. Fortuna, Ł. Ciupiński & K.J. Kurzydłowski Faculty of Materials Science and Engineering, Warsaw University of Technology, Warsaw, Poland

A. Koss, J. Mróz, A. Zatorska & K. Chmielewski Inter-Academy Institute for Conservation and Restoration of Works of Art, Academy of Fine Arts, Warsaw, Poland

J. Marczak, M. Strzelec, A. Rycyk & W. Skrzeczanowski Institute of Optoelectronics, Military University of Technology, Warsaw, Poland

ABSTRACT: A pair of gilded bronze putti, dated at the end of 17th century (Fig. 1) that has been attributed by annotation to the Rome studio of the Dutch sculptor Disqenue, have been part of the decorative elements of Wilanów Palace garden façade. The bronze putti have suffered serious local corrosion, induced by pollution deposits from the atmosphere, beneath the gilding as well as incurring a number of examples of mechanical damage. The paper reports on the development of the ongoing conservation programme, the results of the evaluation of the materials’ compositions and structures. Tests on the influence of laser cleaning in the areas of various deposits and corrosion are included. The major diagnostic investigations include X-ray diffraction analysis, SEM EDS, Raman and LIBS spectroscopy. Set of diagnostic methods allowed full characterization of materials and evaluation of the results of laser cleaning with the use of conventional, Q-switched Nd:YAG laser. However, better cleaning results have been obtained after optimized chemical treatment. Presented results are treated as a starting point for the evaluation of future laser cleaning tests using longer, sub-microsecond laser pulses. 1

INTRODUCTION

The surfaces metal artwork objects exposed to the atmosphere, particularly if copper, copper alloys and gilded or coated, suffer substantially from the effects of atmospheric pollutants. This has occurred on a pair of bronze putti, which have been decorative elements of the façade of the Wilanów Palace in Warsaw for many years (Fig. 1). Solid particles have been deposited on the surfaces and climatic conditions have caused serious local corrosion of the alloy under the gilding (Fig. 2). In addition numerous examples of mechanical damage have been incurred. Laser ablation has been reported as the promising technique for cleaning different kinds of gilded surfaces (Acquaviva et al. 2007, Siano et al. 2003a, Klotzbach et al. 2008, Barrera et al, 2008), because of gold high reflectivity and high damage threshold for practically all wavelengths. Nevertheless, peculiar caution has to be taken due to gilding flaws and losses as well as cracks and corrosion spots in the ground layer (Pantzner et al. 2007).

Figure 1. Pair of bronze putti: 1–putto with torch; 2–putto with laurel.

The most promising efforts were focused on the effects derived from the laser pulse duration in the regions of femtoseconds (Burmeister et al. 2005) and sub-microseconds (Siano et al. 2003b).

111

LACONA8.indb 111

11/10/2010 5:36:48 PM

Figure 2. Examples of environmental impact on putti surfaces: a)–Face of putto with laurel; b)–Wing from putto with torch; c-d)–Laurel fruit and leaf, respectively.

The planned conservation tasks, which include the current work, are as follows: – removal of dirt, dust and corrosion products from all surfaces of the putti, including the removable decorative elements, – neutralization of the corrosion processes (BTA), – camouflage of uncovered minor surface damage and cracks (resulting also from workmanship technology), – repair or replacement of several small elements, gilding and protection of the surface. The first stage presented here determined the experimental basis for further work. The methodology of the research and the experimental results obtained from analysis of the surface deposits, the substrates of the putti and the preliminary laser cleaning tests are described below. The laser used was the portable ReNOVALaser 1 system, see Section 3.1. 2

STUDIES OF THE MATERIALS

The structure and composition of the surface deposits were investigated at various positions on both putti as indicated in Figure 3. The analyses were undertaken in the independent laboratories of MATLAS project partners (http://www.matlas.eu) to permit further standardisation of the methodology and evaluation of different diagnostic methods. In the first stage of the reported studies classical microchemical analyses, SEM EDS, XRD, Raman and LIBS spectroscopy were employed. The results reported in this paper are limited to the most informative and representative.

Figure 3. Indications of measurement points: a) putto with laurel body (without wings)–front view; b) putto with laurel–back side; c) putto with torch body; d) putto with torch shoulder; e) left wing of putto with torch; f) laurel leaf.

The surface structures of the putti were studied using a scanning electron microscope equipped with EDS. Examinations were made in the SE and BSE modes and point analyses of the chemical composition was performed by EDS. The EDS measurements were carried out at 15 kV and 30 kV. The results of the elementary analyses are summarized in Table 1. Examination of the surface of the left wing of the putto with laurel, area 14 in Figure 3, revealed that most of the surface was covered with corrosion products. They were shown, by the images registered in the SE mode, to be accretions (uplifts), while the surrounding flat areas were the retained gilding (Fig. 4). The dark areas registered in the BSE mode correspond to the corrosion products (Fig. 4a). The EDS spectra identified copper, oxygen and sulphur, present at a level of several weight percent, which indicates the presence of copper sulphate. The analysis of the particles embedded in the deposit revealed the presence of alumino-silicates and silicates (quartz). Furthermore, the presence of magnesium (0.6 wt%), potassium (0.2–1 wt%), phosphorus (0.2–0.9 wt%), and areas rich in carbon were detected. Examination of the areas with the retained gilding (Fig. 5) showed numerous surface discontinuities and examples of damage. The two-dimensional structures and compositions of samples were studied using their polished microsections embedded in Meliodent acrylic resin. This enabled the stratiform image of the encrustation, the thickness of the components to be studied and further instrumental analyses to be performed. Interesting examples are shown in Figure 6. The cross-section of a delaminated gilding layer (point 5 in Figure 3c) revealed two metallic coatings—silver and gold. Corrosion products

112

LACONA8.indb 112

11/10/2010 5:36:48 PM

Table 1. No Cu 7 14

Results of EDS analysis [wt%]. O

40–55 25–35 35–45 25–40

S

C

Cl

Fe

10 sev.

– *

1–5 – 1–3 1–4

Ca

Zn

– – 1–1.5 6–11

*—from several to approximately 12 percent dependent on the location. No. 7 & 14 refer to points in Figure 3.

Figure 4. SEM images of the surface of wing of putto with laurel (Figure 3e) in SE mode (left) and BSE mode(right).

Figure 6. Cross sections of samples from point 5 (Fig. 3c): a) presence of Ag and Au layers; b) defect in gilding layer accomplished by a large growth of corrosion products.

Figure 5. Spectrograms of chemical composition at points indicated by numbers (small SEM photographs): a) gilding; b) corrosion.

were located above and below the gilding, substantially growing in the places of flaws (Fig. 6b). Classical microchemical analyses were carried out during the initial studies of the encrustation structure including observations of water smears (microscope Nikon E50i), the reactivity to acids and alkalis as well as ionic micro-crystaloscopic reactions. Ions of Cu2+, S2−, SiO32−, Au3+ and Ag+ were detected, particularly in the black part of the encrustation (point 7 in Fig. 3c and area 14 in Fig. 3e). It could suggest the presence of black copper sulphide. Those results are in accord with data in the Table 1 (point 7). Selected samples were analysed using a scanning electron microscope with an EDS microprobe. Summarizing the results: – The material was an alloy of copper and zinc (17–20 wt%) with the additions of lead,

Figure 7. Lamination of gilding found at point 13 in Figure 3d.

– The grey corrosion products shown in Fig. 6b contain copper, silver, carbon and chlorine, – The red compound in Fig. 6b mainly consisted of copper, carbon, sulfur and oxygen, – The grey corrosion products near the silver layer and above the gilding (grey II in Fig. 6b) contained mainly silver and chlorine and is presumably cerargyrite. Some microsections revealed lamination of the gold layer with red and green corrosion products inside the thin gilding sublayers. Elementary analysis suggested the presence of copper salts, however the expanded composition of microsamples (Fig. 7) necessitates further study.

113

LACONA8.indb 113

11/10/2010 5:36:48 PM

Utilization of a Raman dispersive spectrometer Nicolet Almega enabled several compounds to be identified. Raman spectra were registered for two exciting wavelengths: 532 nm and 780 nm, using high resolution diffraction gratings 2400 lines/mm and 1200 lines/mm. Cuprite Cu2O was identified as the red compound in Figure 6, and the grey salts were probably silver salts. Bands near 235 cm-1 indicate the presence of Ag2O. Figure 8a shows the Raman spectrum of a green copper salt, which has been identified as antlerite Cu3SO4(OH)4 (Fig. 8b), which was found above and below the metallic layers in Figure 6a. Parallel LIBS measurements confirmed most of these results and gave two more important indications. The first concerns the structure of the putti figures—different parts were made of slightly different copper alloys, which can suggest a secondary origin of some decorative elements. The second is probably related to previous conservation procedures. Figure 9b shows the presence of barium,

Figure 8. Raman spectrum of green corrosion sample (a) and standard antlerite spectrum (b).

Figure 9. Identification of overpainting at the laurel branch: a) laurel leaf—place of LIBS measurement; b) LIBS spectrum.

strontium and titanium lines in the LIBS spectrum which indicates a possible secondary painting of the laurel branch (laurel leaf in Fig. 9a) with a mixture of barium and strontium yellow with titanium white. 3 3.1

EVALUATION OF PUTTI CLEANING Laser cleaning

The testing of laser cleaning was carried out to establish the feasibility of precise diagnosing the final appearance of the surface after treatment. A small ReNOVALaser 1 Q-switched system operating at 1064 nm was used (Koss et al. 2008). This system delivers a maximum of 120 mJ per pulse with a variable repetition frequency from 1 to 20 Hz. The pulse width was approximately 6 ns and the beam profile had a top-hat energy distribution at the output of the optical fiber beam delivery system. Testing was carried out on the external side of the right wing of the putto with the laurel. The locations of the treatment and laser cleaning methods are shown in Figure 10. Cleaning with minimum fluence means the application of a laser energy density just above the predetermined threshold of encrustation ablation (0.18 J/cm2—“wet” and 0.45 J/cm2—“dry” laser cleaning in Figure 10). If energy above the minimum fluence is used, i.e. without specific limits, it should be self-limiting and once the surface is clean the energy should be reflected from the cleaned surface and the laser ablation rate drops instantly (so called self-limiting of laser cleaning). Laser fluence was not measured in this case. The evaluation of the cleaning tests was carried out using a scanning electron microscope with EDS spectrometer. The EDS measurements were performed at 15 kV of acceleration voltage.

Figure 10. Diagram of laser cleaning areas at the right wing of putto with laurel.

114

LACONA8.indb 114

11/10/2010 5:36:49 PM

Figure 11. Example of SEM EDS investigations of putto wing surface after laser treatment with minimum fluence: a–b) and d–e) area 1 (Fig. 10); c) and f) area 7 (Fig. 10).

Figure 12. High resolution image of melted fragment of area 11 in Figure 10. Laser fluence—0.5 J/cm2.

Microscopic observations of all areas enabled two characteristic morphologies of the cleaned surface to be distinguished. The first includes areas numbered 1, 2, 7, 8, 11 and 12 in Fig. 11. In all location the laser cleaning revealed the gilding or the putto surface. Part of areas revealed the smooth surface of the gilding (Fig. 11d), composed of gold with up to fifteen percent of silver and copper. In areas covered with droplets (Fig. 11e) mainly copper was detected. It probably indicates melting of the top surface layer down to the substrate, exemplifying agreement with theoretical results (Marczak et al. in press). The unveiled areas also contained numerous cracks, discontinuities and corrosion cavities. At the laser cleaned surface areas a residue of corrosion products was present. Moreover, at the uncovered surface cracks, discontinuities and pits were observed. Inside the pits sliver with copper and oxygen additions was detected (Fig. 11f). This is in good agreement with the micro-fragments

Figure 13. SEM image (a) and composition (b) of melted fragment of area 6 (Fig. 10).

of gilding presented in Figure 6. Areas numbered 2 and 11 were locally melted, which is shown in Figure 12. Much worse results were obtained for unlimited laser fluence on areas 3, 5 and 6 in Figure 10. In these locations the laser treatment uncovered the gilding or the surface, however the surface of the gilding was re-melted. The strongest melting occurred in area 6 (Fig. 13), where the formed

115

LACONA8.indb 115

11/10/2010 5:36:50 PM

Figure 14. a–c) Photographs of putto wing after conventional chemical cleaning; e) topography of representative surface fragment in SEM BSE mode; f–g) high resolution images of gilding damages; h–j) results of SEM EDS analysis at points indicated in e).

layer consisted of gold and silver (50:50). The least damage was observed at area 3, where only local melting occurred. In areas 3 and 5 the chemical composition of the uncovered gilding was a gold alloy with up to fifteen percent of copper and silver. 3.2

Conventional cleaning

Chemical conventional cleaning of selected putto wing was performed using a bath of 10% aqueous ammonia solution followed by a bath distilled water. The results are shown in Figure 14. The white areas in the SEM image in BSE mode (Fig. 14e) correspond to the preserved gilding, which is confirmed by an EDS measurement (Fig. 14h). The grey areas uncovered after corrosion removal, revealed flat areas and pits (probably remains of cavities). Pits were frequently filled with encrustation particles (Fig. 14j). The preliminary analyses have indicated that the conventional chemical method has a slight advantage over the non-optimized laser cleaning discussed earlier particularly when the value of laser fluence was much higher than the encrustation ablation threshold.

4

CONCLUSIONS

The range of material science and optoelectronics diagnostic methods enabled full characterization of the original materials and the encrustations to be made and the results of preliminary tests using laser and traditional, chemical cleaning of the two gilded putti to be evaluated. The analysis of the strata in the surface layers, achieved using microfragments from the putti parts, was similar to that reported in (Siano et al. 2003b) except for the additional silver layer reported here. The encrustation was variably distributed on the gilding surface and contained different chemical compounds, mainly copper and silver oxides and sulphides. In the comparison of cleaning methods, better results were obtained with using an aqueous ammonia solution (5–10%) for washing the items, particularly the inside corrosion cavities in the alloy. Chemical baths are simple and cheap to use as long as only small objects are involved. However, laser cleaning was used only with a conventional shortpulse, Q-switched Nd:YAG laser. In the near-future tasks to be undertaken in the MATLAS project anticipate the development of a sub-microsecond

116

LACONA8.indb 116

11/10/2010 5:36:53 PM

pulse Nd:YAG laser, which would be much more effective and safe for the cleaning of metallic materials (Salimbeni 2006, Marczak et al. 2009). ACKNOWLEDGEMENTS Presented work is carried out in the frames of EEA Financial Mechanism/Norwegian Financial Mechanism grant MATLAS “Advanced methods of materials engineering in diagnostics of art works after renovation by means of shaped, high-energy laser radiation pulses”, contract no. PL0259-GAE00129-E-VI-EEA FM. REFERENCES Acquaviva, S. et al. 2007. Laser cleaning of gilded wood: A comparative study of colour variations induced by irradiation at different wavelengths. Applied Surface Science 253: 7715. Barrera, M. et al. 2008. Application of Ion Beam Analysis (IBA) techniques for the assessment of laser c1eaning on gilded copper. In M. Castillejioet al. (eds), Lasers in the Conservation of Artworks, LACONA VII. Taylor & Francis Group, London: 323. Burmester, T. et al. 2005. Femtosecond laser cleaning of metallic cultural heritage and antique artworks. In K. Dickmann, C. Fotakis, J.F. Asmus (eds), Lasers in the Conservation of Artworks LACONA V. Springer Proc. in Physics 100: 61.

Klotzbach, U. et al. 2008. Potential and limitations of laser technology in restoration of metallic objects of art and cultural heritage Materials and Corrosion 59: 220. Koss, A. et al. 2008. Laser cleaning of set of 18th century ivory statues of Twelve Apostles. In P. Moreno, M. Castillejo, J. Ruiz, R. Radvan, M. Oujja (eds), Lasers in the conservation of artworks LACONA VII. Taylor & Francis Ltd, 162. Marczak, J. et al. 2009. Set of advanced laser cleaning heads and systems. Proc. SPIE 7391: 7391. Marczak, J. et al. Numerical modeling of laser-matter interaction in the region of “low” laser parameters. Applied Physics A. In press. Panzner, M. et al. 2007. Laser Cleaning of Gildings. In J. Nimmrichter, W. Kautek, M. Schreiner (eds), Lasers in the Conservation of Artworks LACONA VI. Springer Proc. in Physics 116: 21. Salimbeni, R. 2006. Laser techniques for conservation of artworks. Archeometriai Műhely 2006/1: 34–40. Siano, S. et al. 2003a. The Santi Quattro Coronati by Nanni di Banco: cleaning of the gilded decorations. Journal of Cultural Heritage 4: 123. Siano, S. et al. 2003b. Laser cleaning methodology for the preservation of the Porta del Paradiso by Lorenzo Ghiberti. Journal of Cultural Heritage 4: 140.

117

LACONA8.indb 117

11/10/2010 5:36:53 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Comparative studies: Cleaning results of short pulsed Nd:YAG vs. fibre J. Hildenhagen & K. Dickmann Laser Center (LFM), Münster University of Applied Sciences, Steinfurt, Germany

ABSTRACT: Compact fiber lasers with only a few watts output power and pulse duration at ns-scale generate sufficient peak power for laser cleaning applications. In combination with an optical scanner these novel lasers may become an interesting alternative to traditional Nd:YAG cleaning lasers. Small dimension, low-maintenance, rugged construction and low initial costs make these laser systems predestinated for versatile applications. The following first comparative studies will show whether cultural and historical objects are adapted for this application as well. It has turned out that positive results could be obtained by applying this new technique on flat surfaces with less thermal sensitivities. Furthermore especially homogenous areas could be cleaned more smoothly than by using a hand guided Nd:YAG Laser. However, the linear scan routine often generates a defined line pattern onto the cleaned surface, which can be seen either directly by naked eye or by microscope. Polychrome surfaces react as sensitive as to Nd:YAG Laser irradiation. 1

INTRODUCTION

Recent short pulse fiber laser sources represent a compact and reliable tool for various applications (s. Fig. 1). They can be applied for cleaning applications if a scanner system is utilized. In comparison to conventional Nd:YAG-laser cleaning systems a completely different handling system and processing scheme is applied. Thus, contrary to the application of a large spot diameter (8 mm) of manually operated cleaning systems with 10 pulses per second, a tightly focused laser beam (50 μm) with thousands of pulses per second is used for the case of fiber laser sources and scanner technology. For this study a stationary laser scanner system with two additional linear axis was applied which enables reproducible cleaning results (s. Fig. 2). Hereby only flat samples could be considered. A traditional q-switched Nd:YAG Laser cleaning system from Thales (SAGA 220/10) with articulated arm was used as reference for this study. Both systems emitted IR-radiation at λ = 1064 nm and were respectively used at optimal parameters and comparable energy densities. The pulse duration of the Nd:YAG Laser is fixed at 9 ns but can be varied for the SPI fiber laser system via the waveform between 9 and 200 ns. However, short pulse durations cause a collapse of the pulse energy (s. Fig. 3). Increasing the repetition rate stabilizes the average output power but influences the interaction process between laser irradiation and sample.

Figure 1. Short pulsed fiber laser SPI G3 with 20 W (Source: SPI Lasers).

Figure 2. Experimental setup for laser cleaning application with short pulsed fiber laser.

119

LACONA8.indb 119

11/10/2010 5:36:53 PM

Figure 3. Different pulse shapes of the SPI G3 (Source: SPI Lasers).

Figure 4. Comparison of laser cleaning using (an) 8 mm spot (Nd:YAG, 10 Hz, left) and (a) 40 μm spot (Fiber Laser, 25 kHz).

In case of applying a fiber laser a several thousand times more laser pulses will be needed for cleaning an equal area than by using a Nd:YAG Laser (s. Fig. 4). Due to the high repetition rate within kHz-scale this amount of laser pulses is generated in almost the same time as the few pulses of a Nd:YAG Laser. Thus the cleaning rate is comparable for both systems. However, for the handling there are significant differences between both laser systems, since the fast pulsed fiber laser radiation can only be guided in defined lines by using a scanner optic. Selection of the waveform and other laser parameters such as output power and repetition rate are controlled via software of the scanner system. The software also controls other scanner values like spot velocity, filling pattern and scan geometries. Based on these conditions several studies on different materials were carried out and compared to the cleaning results of the manual articulated Nd:YAG Laser. 2

MARBEL

On weathered marble both laser systems achieved nearly identical cleaning results at a cleaning rate

Figure 5. Weathered marble surface cleaned by Nd: YAG Laser (see above) and by fiber laser (see below).

of ∼ 0,2 m2/h. The visual results convinced in both cases; there were no significant differences for the L*a*b-spectra. Only SEM-pictures revealed a higher degree of purification for the Nd:YAG (s. Fig. 5). In order to obtain a smooth cleaning result using the scanner guided fiber laser, the deviation out of plane of the substrate has to be within ±2 mm. Otherwise the variation of energy density let to inhomogeneous results. 3

PARCHMENT

Aged parchment shows a considerable sensitivity for the chosen fiber laser parameters. Thermal reaction and blackening of the organic material occurred very easily. Only the combination of low energy density (<0,2 mJ/cm2) and multiple scan cycles leads to satisfying results. An area of 0,1 m2/h could be cleaned by fiber laser, this means only one-fifth of cleaning capacity of the Nd:YAG Laser was achieved. Variation of the waveform shows a higher cleaning effect at 65 ns pulse duration than at 9 ns or

120

LACONA8.indb 120

11/10/2010 5:36:54 PM

Figure 8. Line pattern (ghost lines) on derusted steel surface, caused by line scanning with fiber laser.

Figure 6. Surface of parchment, cleaned by fibre laser; variation of pulse shape and pulse duration.

of the surface causes objectionable side effects in term of a slight line pattern. No parameter combination could be found in order to remove the corrosion layer completely without leaving a line pattern on the surface. This effect is caused by a fast melting and solidification process within the area of the laser spot. In combination with the scanner routine “ghost lines” were generated (s. Fig. 8). In some cases this line structure is even visible to the naked eye. Reason for local melting of the surface is the deposit of remaining radiation energy within a short interaction time. At LACONA 8 an additional paper of the authors gives a more detailed report about this effect. 5

Figure 7. Laser.

Surface of parchment, cleaned by Nd:YAG-

200 ns (s. Fig. 6), respectively. However, the remaining laser parameters were kept constantly during that test. This means that a further improvement of the cleaning results is promising by parameter optimization. As well in this case the taken SEM-Pictures demonstrate a higher cleaning level by applying a Nd:YAG Laser (s. Fig. 7). 4

STEEL

Corroded steel could also be cleaned by using both laser systems. But in case of fibre laser the scanning

PAINTED SURFACES

As well known, in general colors react very sensitive to IR-Irradiation, particularly organic pigments underlie a chemical transition which is expressed by discoloration. For example the removal of dyed varnish (sturgeon glue with umbra) is neither possible by fiber laser nor by Nd:YAG-Laser in the majority of cases. During this study solely titanium white turned out to be suitable to only a limited extent but also on this material occurred fine line pattern which can be detected by microscope (s. Fig. 9). This pattern is not induced by a thermal reaction but due to a suboptimal overlap of the single lines during the scanning process. An improvement is hindered because of Gaussian energy distribution within the laser beam. More promising findings were documented by another work group applying a mobile short pulse fiber laser on site of restoration for cleaning for cleaning mural paintings in an Egyptian chamber.

121

LACONA8.indb 121

11/10/2010 5:36:55 PM

Figure 9. Removal of dyed varnish on titanium white, Ep = 0,8 J/cm2 (left) and 0,6 J/cm2 (right) by fiber.

6

SUMMARY

The applied fiber laser with 20 W output power offers an enormous potential. Compact size and modest requirements relating to electrical power input, cooling and service open a wide range of applications. However, in the field of cleaning the given advantages of the fiber laser can only be utilized in combination with a scanner system, on the other hand this makes a flexible application more difficult. The high level of beam demagnification let to a small depth of focus which allows only minimal distance variations without shifting results. Often the generation of line pattern on the cleaned

surface could not be avoided. However in many case these structures are only visible with the help of a microscope. On cultural objects with a higher thermal resistance like marble, pleasant cleaning results could be achieved with the same efficiency like Nd:YAG-Laser at comparable output power. More sensitive samples like parchment need laser parameter with less energy density and many repetitions whereas the cleaning efficiency decreases significant. In summary the fiber laser generates more heat input as the application of Nd:YAG-Laser. In the field of restoration the fiber laser at its current stage of development is an interesting addition for several applications but can not be seen as an adequate alternative for Nd:YAG-Laser. However, this cleaning method offers a high potential of further developments and demonstrates that the field of laser cleaning is not utilized yet. REFERENCES Chappé, M. et al., J. Cult. Heritage 4 (2003): 264–270. Hildenhage, J. & Dickmann, K., Restauro 7, 466–470, Callwaey, Munich 2009. Hildenhagen, J., et al., LACONA V Proceedings, 297–302, Springer, Berlin 2005. Thorsten Naeser: Ein Laser für Neferhotep, Abenteuer Archaeologie 3, S. 72–73, Verlag Spektrum der Wissenschaft, Heidelberg 2006.

122

LACONA8.indb 122

11/10/2010 5:36:55 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Laser cleaning of iron: Surface appearance and re-corrosion of model systems C. Korenberg & A.M. Baldwin The British Museum, London, UK

ABSTRACT: Model iron coupons corroded to different extents were cleaned mechanically or using a Q-switched Nd:YAG laser at 1064 nm and left in ambient conditions for more than 19 months to re-corrode, together with an un-corroded control. For lightly corroded coupons, laser cleaning was judged to be the most effective cleaning technique and re-corrosion of the laser cleaned and mechanically cleaned coupons occurred to a similar extent. For more heavily corroded coupons, the best cleaning results were obtained by combining mechanical and laser cleaning. Coupons cleaned using a combination of cleaning methods re-corroded to a slightly greater extent than the mechanically cleaned coupons, but the control coupon appeared most corroded suggesting that residual corrosion decreases the degree of re-corrosion. This study, based on visual and microscopic assessment, suggests that laser cleaning may be a suitable conservation technique for cleaning historical corroded iron. 1

of laser cleaning on re-corrosion. The results of the experiments are presented below.

INTRODUCTION

In conservation active corrosion products are often removed from the surface of historical iron artefacts using wire wool and a solvent. This method is not entirely satisfactory as it can alter the surface of the metal –either polishing textured surfaces or etching smooth surfaces– and may leave some corrosion, especially on rough surfaces. Laser cleaning has gained considerable success as a valuable method of conservation and its potential use for cleaning metal artefacts has been the subject of many publications since 1995. Previous work on the laser cleaning of non-archaeological corroded iron (Korenberg and Baldwin 2007) has shown that short pulses at 1064 nm were the most effective at removing corrosion without affecting the metal substrate. However, only spot tests on very small areas (typically 2 mm diameter) were performed and the longer-term effects of laser cleaning were not assessed. The aim of the present study was to explore the practical use of the laser on corroded iron and the long-term effects of laser cleaning in order to assess, based on visual and microscopic appearance, whether laser cleaning is a suitable conservation technique for cleaning corroded iron. Iron coupons were corroded to different extents, either naturally or artificially, and cleaned mechanically, using a Q-switched Nd:YAG laser at 1064 nm and with a combination of both methods. Sets of coupons were then left to corrode in the laboratory for more than 19 months to test the long-term effects

2

COUPONS

The coupons used for the experiment were rolled sheets of iron (99.5% purity with traces of manganese, silicon, carbon, phosphorus and sulphur). The sheets were cut in half to give a coupon size of approximately 25 × 50 × 1 mm. A set of coupons (‘set I’) were corroded artificially for one week in a chamber humidified to 75% RH (±5%) using a saturated aqueous sodium chloride solution. This produced a lightly corroded surface, with no pitting. Two sets of coupons (‘sets II and III’) were corroded artificially in the same conditions as described above for four weeks, producing an even layer of corrosion with shallow pitting of the metal surface. Finally, a fourth set of coupons (‘set IV’) were left to corrode naturally in the laboratory environment for approximately one year producing a thin, “patchy” layer of corrosion. Photographs of the different coupons are shown in Figure 1. 3 3.1

CLEANING TESTS Mechanical cleaning

A coupon from each set was cleaned mechanically using wire wool and white spirit. Using this method it was possible to remove most of the corrosion on the lightly corroded coupon, but some corrosion

123

LACONA8.indb 123

11/10/2010 5:36:55 PM

Figure 1. From left to right: Lightly corroded coupon, heavily corroded coupon and naturally corroded coupon.

was left in pits on the heavily and naturally corroded coupons (see Figure 2). When examined using Scanning Electron Microscopy (SEM), scratches could be seen on the surface of the mechanically cleaned coupons (see Figure 3) and the overall appearance was quite different from that of an un-corroded coupon (see Figure 4). Also, the mechanically cleaned coupons were darker in colour than un-corroded coupons with a slight brownish tint when viewed under a raking light. 3.2

Figure 2. Appearance of coupons after mechanical cleaning: Heavily corroded coupon (left) and naturally corroded coupon (right).

Laser cleaning

Laser cleaning was carried out on one coupon from each set with a Lynton Phoenix Q-switched Nd:YAG laser using the 1064 nm wavelength at different fluences (see Table 1). This laser emits 10 ns pulses. Water was applied on the surface of the coupons as the use of a wetting agent has been reported to increase the cleaning ability of the laser compared to dry cleaning (Korenberg and Baldwin 2007, Koh et al. 2006, Dickmann et al. 2005, Koh & Sarady 2003). All the corrosion was removed (as judged by visual assessment) from the lightly corroded coupons following laser cleaning at 0.2 J/cm2 and no surface melting was observed using SEM. Using the laser did not result in the brownish tint seen on the mechanically cleaned coupon. Results of laser cleaning on sets II and III (heavily corroded coupons) were less satisfactory as corrosion was still present in the pits after cleaning as can be seen in Figure 5. Applying further laser pulses to these areas led to blackening of the corrosion and melting of the adjacent metal surface. The darkening of iron corrosion products has been reported by several authors (Korenberg and Baldwin 2007, Dickmann et al. 2005, Koh & Sarady 2005, Koh & Sarady 2003, Cottam & Emmony 1999) and is thought to be caused by the decomposition of iron oxyhydroxide (FeOOH) to

Figure 3. SEM image of the surface of a heavily corroded coupon that was mechanically cleaned, showing microscopic scratches caused by the mechanical cleaning with wire wool (length of the scale marker: 100 microns).

Figure 4. SEM image of the surface of an uncorroded coupon (length of the scale marker: 100 microns).

124

LACONA8.indb 124

11/10/2010 5:36:55 PM

Table 1.

mechanical cleaning, laser cleaning was less even, but the colour of the laser cleaned coupons was closer to the original appearance than the mechanically cleaned coupons with no brownish tint. Also, at a microscopic level, the appearance of the laser cleaned surface was closer to that of un-corroded coupons than the mechanically cleaned surface.

Fluences used to clean the coupons.

Coupons

Fluence (J/cm2)

Lightly corroded (set I) Heavily corroded (set II) Heavily corroded (set III) Naturally corroded (set IV)

0.20 0.34 0.68 0.68

3.3

Combined cleaning

Coupons from sets II, III and IV were cleaned by a combination of mechanical and laser cleaning: they were first lightly mechanically cleaned to remove the bulk of the corrosion, taking care not to scratch the surface, and then laser cleaning was used to remove the remaining corrosion. Some corrosion was left on the coupons using this method (see Figure 8), but more corrosion appeared to be removed than when using laser cleaning or mechanical cleaning alone. As in the case of the laser cleaned coupons,

Figure 5. Appearance of heavily corroded coupons after laser cleaning: Coupon cleaned at 0.34 J/cm2 (left) and coupon cleaned at 0.68 J/cm2 (right).

Figure 7. Appearance of naturally corroded coupon after laser cleaning at 0.68 J/cm2.

Figure 6. SEM image of the surface of a heavily corroded coupons laser cleaned at 0.68 J/cm2. Very slight melting occurred as indicated by the presence of small white spheres (field of view: 100 microns).

maghemite (γ-Fe2O3) or magnetite (Fe3O4) under the effect of heat. More corrosion was removed at 0.68 J/cm2 than at 0.34 J/cm2 (based on visual appearance), but there appeared to be some very slight melting at a microscopic level on the coupons laser cleaned at the higher fluence, as seen using SEM (see Figure 6). Some corrosion also remained on the naturally corroded coupon following laser cleaning at 0.68 J/cm2 (Figure 7) and very slight melting was observed using SEM. Compared to

Figure 8. Appearance of coupons after combined mechanical and laser cleaning. From left to right: Heavily corroded coupon cleaned at 0.34 J/cm2, heavily corroded coupon cleaned at 0.68 J/cm2 and naturally corroded coupon cleaned at 0.68 J/cm2 (note that it was not possible to remove corrosion from two large pits on this coupon).

125

LACONA8.indb 125

11/10/2010 5:36:56 PM

the colour and the appearance at a microscopic level of the cleaned metal was very close to that of the un-corroded original surface. More corrosion was removed at 0.68 J/cm2 than at 0.34 J/cm2, but there appeared to be some very slight melting at a microscopic level on the coupon cleaned at the higher fluence, as seen using SEM. Some corrosion remained on the naturally corroded coupon after cleaning but less than on the equivalent solely laser cleaned coupon and very slight melting was observed using SEM. 4

RE-CORROSION TESTS

After cleaning, the coupons were left in the laboratory, where the relative humidity varied between 15 and 85%, and the degree of re-corrosion was monitored visually. Set I (lightly corroded coupons) and set II (heavily corroded coupons) were left for 24 months, while sets III (heavily corroded coupons) and IV (naturally corroded coupons) were left for 19 months. For sets III and IV, an un-corroded coupon was included as a control sample. 4.1

Lightly corroded coupons (set I)

Small spots of corrosion formed more rapidly on the laser cleaned coupon than on the mechanically cleaned coupon, as shown in Figure 9 (after nine months), but the two coupons appeared to be corroded to a similar extent after 24 months (Figure 10). 4.2

Figure 10. Appearance of set I after 24 months: Laser cleaned coupon (left) and mechanically cleaned coupon (right).

Heavily corroded coupons (set II)

Figure 11. Appearance of set II at the start of the experiment. From left to right: Coupon cleaned using combined methods, mechanically cleaned coupon and laser cleaned coupon.

The re-corrosion on the coupons from set II varied according to the cleaning method. On the mechanically cleaned coupon the re-corrosion was

Figure 12. Appearance of set II after 9 months. From left to right: Coupon cleaned using combined methods, mechanically cleaned coupon and laser cleaned coupon.

Figure 9. Appearance of set I after 9 months: Laser cleaned coupon (left) and mechanically cleaned coupon (right).

uniform across the surface, whilst the formation was less evenly distributed on the other coupons (Figures 11 and 12). The unevenness of the corrosion was more pronounced on the laser cleaned coupon. After 24 months all the coupons from set II were heavily corroded, but the mechanically cleaned coupon appeared to be slightly less corroded (Figure 13).

126

LACONA8.indb 126

11/10/2010 5:36:57 PM

4.3

Heavily corroded coupons (set III)

It was suspected that the patchy appearance of the re-corrosion on the laser cleaned coupon and the coupon cleaned using combined methods of set II might be due to residual corrosion remaining on the edges of the coupons. Therefore, the edges of the coupons for set III were mechanically cleaned prior to the re-corrosion experiment. The re-corrosion on the coupon cleaned using the combined method was relatively even, although formation of corrosion on the laser cleaned coupon was still patchy (Figures 14–16). At the end of the experiment, the mechanically cleaned coupon appeared slightly less corroded than the coupon cleaned using the combined methods (Figure 16). However, the control

coupon appeared to be the most corroded by far, perhaps suggesting that the presence of residual corrosion reduces the extent of re-corrosion. This has been noticed by other researchers studying the corrosion of iron (Wang 2007) but without investigating the surface composition and chemistry of the coupons before and after the re-corrosion experiment, a firm conclusion cannot be drawn. 4.4 Naturally corroded coupons (set IV) There was little difference in the appearance of the coupons as they corroded in set IV and at the end the experiment all the coupons appeared to be corroded to a similar extent (Figures 17–19). However, as with set III, the control coupon appeared to be corroded to a greater extent than the test coupons.

Figure 16. Appearance of set III after 24 months. From left to right: Mechanically cleaned coupon, laser cleaned coupon, coupon cleaned using combined methods and control coupon.

Figure 13. Appearance of set II after 24 months. From left to right: Coupon cleaned using combined methods, mechanically cleaned coupon and laser cleaned coupon.

Figure 14. Appearance of set III at the start of the experiment. From left to right: Mechanically cleaned coupon, laser cleaned coupon, coupon cleaned using combined methods and control coupon.

Figure 15. Appearance of set III after 4 months. From left to right: mechanically cleaned coupon, laser cleaned coupon, coupon cleaned using combined methods and control coupon.

Figure 17. Appearance of set IV at the start of the experiment. From left to right: Mechanically cleaned coupon, laser cleaned coupon, coupon cleaned using combined methods and control coupon.

Figure 18. Appearance of set IV after 4 months. From left to right: Mechanically cleaned coupon, laser cleaned coupon, coupon cleaned using combined methods and control coupon. The two pits of corrosion remaining on the combined cleaned coupon were covered using lacquer and microcrystalline wax to restrict re-corrosion from these areas.

127

LACONA8.indb 127

11/10/2010 5:36:58 PM

after the combined treatment. However, without investigating the surface composition and chemistry of the coupons before and after the re-corrosion experiment, a firm conclusion cannot be drawn and this would need to be investigated further. 6 Figure 19. Appearance of set IV after 19 months. From left to right: Mechanically cleaned coupon, laser cleaned coupon, coupon cleaned using combined methods and control coupon.

5

DISCUSSION

The laser cleaning appears to have been very successful on the lightly (artificially) corroded coupons (set I) based on visual and microscopic assessment. All the corrosion could be removed at a low fluence without surface melting and laser cleaning removed more corrosion than the mechanical cleaning employed. In the re-corrosion experiment, the laser cleaned coupon and the mechanically cleaned coupon appeared to have corroded to a similar extent after 24 months. Therefore, laser cleaning seems to be a suitable conservation technique for lightly artificially corroded iron objects and may therefore be suitable for lightly corroded historical iron artefacts. For the other coupons (sets II–IV), the best visual results were obtained using a combination of light mechanical cleaning followed by laser cleaning at 0.68 J/cm2. More corrosion was removed than using either mechanical cleaning or laser cleaning alone. However, laser cleaning caused some very slight melting at a microscopic level as seen at high magnification using SEM. Micromelting is obviously undesirable, but conventional cleaning using mechanical methods was found to alter the surface of the metal to a greater extent at a microscopic level. At the end of the re-corrosion tests, the coupons from set IV (naturally corroded coupons) appeared to have re-corroded to the same extent regardless of the cleaning procedure. For the artificially corroded sets (sets II and III), the mechanically cleaned coupons appeared slightly less corroded than the coupons cleaned using combined methods. Nevertheless, at the end of the re-corrosion experiment, all of the cleaned coupons appeared much less corroded than the control coupon, suggesting that residual corrosion decreased the degree of re-corrosion. Thus, it is possible that coupons cleaned using the combined methods corroded to a greater extent than the mechanically cleaned coupons probably because less corrosion was left

CONCLUSIONS

In this experiment, laser cleaning at a fluence of 0.2 J/cm2 was very successful on the lightly (artificially) corroded coupons in terms of surface appearance and re-corrosion. For the cleaning of heavily corroded iron, a combination of mechanical and laser cleaning gave a superior visual appearance when compared to mechanical cleaning. Coupons cleaned using combined methods re-corroded to a slightly greater extent than the mechanically cleaned coupons, but the control coupon appeared most corroded suggesting that residual corrosion decreases the degree of re-corrosion. This study suggests that laser cleaning may be a suitable conservation technique for cleaning historical corroded iron, if care is taken to use the lowest suitable fluence and if used in combination with mechanical methods for heavily corroded iron. REFERENCES Cottam, C.A. & Emmony, D.C. 1999. TEA-CO2 laser surface processing of corroded metals. Corrosion Science 41: 1529–1538. Dickmann, K., Hildenhagen, J., Studer, J. & Müsch, E. 2005. Archaeological ironwork: removal of corrosion layers by Nd:YAG laser. In K. Dickmann, C. Fotakis & J.F. Asmus (eds), Lasers in the conservation of artworks: LACONA V proceedings, Osnabrück, 15–18 Sept. 2003. Berlin: Springer. Koh, Y.S. & Sarady, I. 2003. Cleaning of corroded iron artefact using pulsed TEA CO2 and Nd:YAG lasers. Journal of Cultural Heritage 4: 129–133. Koh, Y. & Sarady, I. 2005. Surface cleaning of iron artefacts by lasers. In K. Dickmann, C. Fotakis & J.F. Asmus (eds), Lasers in the conservation of artworks: LACONA V proceedings, Osnabrück, 15–18 Sept. 2003. Berlin: Springer. Korenberg, C. & Baldwin, A. 2007. Investigating and optimising the laser cleaning of corroded iron. Lasers in the conservation of artworks: LACONA VII proceedings, Madrid, 17–21 Sept. 2007, in press. Wang, Q., Personal communication, 10 June 2007.

128

LACONA8.indb 128

11/10/2010 5:37:01 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Reversion of darkened red lead-containing wall paintings by means of cw-laser irradiation: In situ tests and first application S. Aze Centre Interdisciplinaire de Nanoscience de Marseille et Centre Interrégional de Conservation et Restauration du Patrimoine, Marseille, France

J.-M. Vallet Centre Interrégional de Conservation et Restauration du Patrimoine, Marseille, France

V. Detalle Laboratoire de Recherche des Monuments Historiques, Champs sur Marne, France

O. Grauby Centre Interdisciplinaire de Nanoscience de Marseille, Marseille, France

ABSTRACT: Final developments of an innovative restoration technique for the reversion of darkened red lead pigment in mural paintings are presented. The technique is based on the photo-thermal reduction of black lead dioxide (plattnerite, β-PbO2) into minium (Pb3O4), the main component of the original red lead pigment. Local heating of the darkened pictorial layer is produced by continuous-wave laser irradiation. Optimization of the reversion process is achieved through the choice of suitable irradiation parameters, including laser wavelength, beam profile and power, and irradiation time. Tests on both plattnerite samples and naturally aged red lead paint samples show the high efficiency of fiber-optics diode system emitting at 811 nm. The first in situ treatment of a mural painting containing a darkened red lead pigment is presented. 1

INTRODUCTION

1.1 Red lead darkening in paintings Red lead pigment, one of the first artificial pigments for painted artworks, is known to undergo chromatic alterations during ageing (Aze et al. 2008). Such alterations particularly affect wall paintings, where red lead darkening originates from the pigment alteration into plattnerite (β-PbO2), a mineral form of black lead dioxide (Aze et al. 2007a). In particular, red lead transformation into plattnerite was observed in medieval frescoes (Bollingtoft & Christensen 1993; Daniilia et al. 2000; Daniilia & Minopoulou 2009), but also in Antique Asian paintings (Piqué 1997). Recent investigations of red lead alteration process point out the role of dissolved pollutants, such as carbon dioxide and sulfur dioxide (Aze et al. 2006; Daniilia et al. 2008). Minium (Pb3O4), the main component of red lead, is most likely disproportionated into a mixture of plattnerite and lead carbonate (cerusite, PbCO3) and/or lead sulfate (anglesite, PbSO4). Such secondary phases are occasionally detected on discolored red lead (Kotulanova et al. 2009), but may more often be leached during ageing due to a significant solubility

(Aze et al. 2005). Resulting darkened red lead paint layer may therefore consist of a thick black layer mainly constituted by plattnerite particles. Due to the complexity of Lead/Oxygen chemistry, no specific method had been developed for the restoration of darkened red lead in paintings. Lead, indeed, can have both +2 and +4 oxidation states. A large number of both stoichiometric and non-stoichiometric Lead oxides occur in nature, including litharge (α-PbO, tetragonal), massicot (β-PbO, orthorhombic), minium (Pb3O4) and plattnerite (β-PbO2) (Fig. 1).

Figure 1. Lead-oxygen composition diagram showing the main defined lead oxides.

129

LACONA8.indb 129

11/10/2010 5:37:01 PM

However, plattnerite reversion into minium may be achieved through a thermal treatment over 375°C (Burgio et al., 2001). Plattnerite is partially reduced while oxygen is emitted, leading to crystallized minium particles. These latter are stable over the 375°C–580°C temperature range, further heating producing either litharge (T > 580°C) or massicot (T > 650°C) (Gavrichev et al. 2008). 1.2

Photo-thermal reversion of plattnerite into minium

Such a process may be obtained through photothermal effect using continuous-wave (cw) laser irradiation. Optical absorption of coherent photons by plattnerite particles, followed by heat diffusion through the plattnerite layer, induces a local temperature rise. Depending on the laser irradiation parameters, plattnerite reduction into pure minium may be achieved, yielding no residual plattnerite or lead monoxide. The use of laser beam offers many benefits for such a new application: – Spatial accuracy: Laser beam can be shaped using optical devices, such as mirrors, lenses and diaphragms, so that the interaction size and location can be controlled and monitored. – Flexibility: Irradiation parameters (laser power, irradiation time) can be controlled and – Portability: The use of fiber-optics as an optical guide or the active gain medium, allows in situ treatment of paintings. Preliminary investigation of such a new restoration technique had been presented in Lacona 7 conference (Aze et al. 2007b). Promising results had been obtained using a laboratory Nd:YAG cwlaser. However few amounts of residual plattnerite, together with non-stoichiometric lead oxide of formula Pb2O3.33 were present in treated plattnerite samples. Attempts to optimize the reversion process are presented here, together with first in situ irradiation tests and application on an historical artwork.

2

including optical microscopy, X-Ray Diffraction (XRD), and Micro-Raman Spectroscopy (MRS). Observations and analyses showed the transformation of red lead pigment into plattnerite within a large depth. Plattnerite was found to be the main component of the darkened layer, which was more than 20 micrometers in depth. Few amounts of gypsum were detected, mainly at the paint layer surface. In few cases, Lead II compounds, such as anglesite (PbSO4) and cerusite (PbCO3) were also detected in small amounts (Aze et al. 2007b). 2.2

Laser systems

Optimization of the irradiation parameters was carried out with the aim of obtaining a homogeneous minium layer. Different cw-laser systems have been employed in order to investigate the influence of laser wavelength in the plattnerite reversion process. Visible green source was provided by a Ar+ laser equipment, while near-infrared irradiations at 1064 nm and 811 nm were supplied by Nd:YAG and fiberoptics diode systems, respectively (Table 1). Each laser beam was shaped using a specific set of optical devices, in order to obtain a quasi-parallel beam of few millimeters in diameter. The resulting beam irradiance profiles, obtained from CCD camera measurements, are presented in Figure 2. Table 1.

Characteristics of the laser systems.

Laser source

Wavelength (nm) Supplier

Ar+

488

cw Nd:YAG 1064 Fiber-optics 811 diode system

Model

Max. power

Coherent Innova 10 W 200-10 LILM Experimental 60 W Coherent FAP-S 30 W

EXPERIMENTAL

2.1 Materials: Naturally aged experimental wall paintings Darkened red lead-containing experimental wall paintings were used as standard samples. As detailed in previous papers (Morineau & Stefanaggi 1995; Aze 2005; Aze et al. 2007a), these wall paintings naturally aged during 28 years, leading to red lead alteration into plattnerite. Samples were taken from darkened fresco-like areas and characterized by suitable methods,

Figure 2. Laser irradiance profiles of Ar+, Nd:YAG and fiber-optics diode lasers.

130

LACONA8.indb 130

11/10/2010 5:37:02 PM

2.3

Methodology

Successive irradiation tests were carried out on darkened paint samples using each laser source. Combinations of different laser power, beam size and irradiation time were applied. The effects of the irradiation were initially estimated through microscopic observations of the sample surface. Further characterization of the treated samples, e.g. paint layer stratigraphy and composition, was achieved by microscopic observations, XRD and MRS. 3 3.1

RESULTS Laboratory tests on plattnerite samples

Initial irradiation tests were carried out on plattnerite samples in order to establish peak power density thresholds for plattnerite to minium and minium to massicot reductions, labeled as P1 and P2, respectively (Table 2). A long irradiation time, viz. 60 seconds, was applied to avoid partial thermal reaction. Peak power density was deduced from global power emission values measured by a laser power meter and 2D-beam irradiance profiles. Irradiation by Ar+ laser produced a slight reddening of the plattnerite surface, promptly followed by a yellowing of the red particles, predominantly at the center of the laser spot. No suitable laser power was able to produce a homogeneous reddening of the sample surface. Nd:YAG laser irradiations showed different behavior. Red particles appeared at slightly higher power density (P > 8.6 W.cm−2); yellowing of red particles took place when power was strongly increased (P > 200 W.cm−2). Within this gap, the higher the power was set, the most saturated the red color. Micro-Raman analyses of the red areas confirmed the reversion of plattnerite particles into minium and further reduction into massicot at excessive laser powers (Figure 3). Fiber-optics NIR irradiation required a similar power density to induce plattnerite reddening. Highly saturated and homogeneous red areas were obtained within a large power density range, viz. Table 2. Laser power density thresholds for plattnerite to minium reduction (P1) and minium to massicot reduction (P2). Laser source Ar+ cw Nd:YAG Fiber-optics diode system

Wavelength (nm)

P1 (W.cm−2)

P2 (W.cm−2)

488 1064 811

6.6 8.6 8.1

7.8 200 17

Figure 3. Raman analyses of darkened red leadcontaining paint sample after Nd:YAG laser irradiation, in the original dark (a), yellow (b) and red areas (c), matching with reference spectra of plattnerite, massicot and minium, respectively.

between 10 and 17 W.cm−2., the best results being obtained at 15 W.cm−2. Based on these results, in situ irradiation test were carried out on the experimental wall paintings using the fiber-optics NIR diode laser. The optical head was mounted on a manually controlled XYZ stage (Figure 4). Successive areas were treated at different laser powers and using different scanning speeds. Optimal combination of these parameters was established visually according to color tone and saturation. Orange-red homogeneous areas were obtained at 22 W.cm−2 with a scanning rate of ca. 13 s.cm−1. Subsequent XRD analyses of both original dark brown and final red areas show the effective reversion of plattnerite into minium (Figure 5). Samples taken in both dark brown and red areas were embedded and prepared as thick crosssection. Microscopic observations reveal the transformation of the whole plattnerite layer depth, which had an original thickness of 20 to 50 micrometers (Figure 6). Micro-Raman analyses carried out at various depths confirm the presence of minium, with no residual plattnerite or lead monoxide. No physical alteration of the pictorial layer, such as cracking or scaling, was detected. This reversion technique is then very suitable to fresco and frescolike techniques. 3.2

In situ treatment of an historical mural painting

Based on both laboratory and in situ optimization tests, fiber-optics diode laser irradiation was applied for the treatment of a darkened red lead-containing mural painting (XIXth c.) from

131

LACONA8.indb 131

11/10/2010 5:37:02 PM

media which is used in some techniques of wall paintings because of a too high temperature.

4

Figure 4. In situ treatment of naturally aged experimental wall paintings using the fiber-optics diode laser.

CONCLUSION

Interaction of plattnerite samples with laser beam was found strongly dependent on the laser beam properties, i.e. Laser wavelength, power, and power spatial distribution. No further transformation took place after a short irradiation time of about 5 seconds, independently of laser source. Optimal results were obtained using fiber-optics NIR diode laser at 15 W.cm−2. Plattnerite layer, as thick as 50 micrometers, was reverted into a cohesive, homogenous red layer constituted by minium. In situ application on both experimental and historical mural paintings proved the efficiency of the reversion process. Further investigations are needed in order to have a good approach to the treatment of blackened wall paintings containing also organic media or other pigments sensitive to the blackening transformation (such as lead white) or sensitive to the thermal effects induced by the laser irradiation.

REFERENCES

Figure 5. X-ray diffraction patterns of a sample surface after irradiation using fiber optics diode laser (15 W.cm−2) compared to the XRD pattern of an original darkened area, showing the main peaks of Minium (M) and Plattnerite (P), respectively.

Figure 6. Sample cross-sections of the original darkened paint layer (a) and the red layer after treatment using fiber-optics diode laser (b).

the Chapel of Solomiat (XVth–XVIth c., Ain, France). Reversion of darkened areas was carried out with the aid of a paint restorer after local validation tests. Original red areas were effectively reverted; darkened flesh tone areas, originally composed by red lead and lead white, were reverted into red minium only. As another current problem of this method of reversion is the destruction of organic

Aze S. 2005. Altérations chromatiques des pigments au plomb dans les œuvres du patrimoine- Etude expérimentale des altérations observé sur les peintures muralesPhD thesis, Université de Marseille (France). Aze S., Vallet J.-M., Baronnet A. & Grauby O. 2006. The fading of red lead pigment in wall paintings: tracking the physico-chemical transformations by means of complementary micro-analysis techniques. European Journal of Mineralogy 18 (6): 835–843. Aze S., Vallet J.-M., Baronnet A. & Grauby O. 2007a. Red lead darkening in wall paintings: natural ageing of experimental wall paintings versus artificial ageing tests. European Journal of Mineralogy 19 (6): 883–890. Aze S., Delaporte P., Detalle V., Grauby O., Vallet J.-M. & Baronnet A. 2007b. Towards the restoration of darkened red lead-containing mural paintings: a preliminary study of the β-PbO2 to Pb3O4 reversion by laser irradiation. Lasers in the conservation of artworks: proceedings of the international conference Lacona VII, Madrid, Spain, 17–21 September 2007, 11–13. Aze S., Vallet J.-M., Detalle V., Grauby O. & Baronnet A. 2008. Chromatic alterations of red lead pigments in artworks: a review. Phase Transition 81 (2–3): 145–154. Bollingtoft P. & Christensen M. 1993. Early gothic wall paintings: an investigation of painting techniques and materials of 13th century mural paintings in a Danish village church. In Bridgland, Janet (ed.), ICOM Committee for Conservation 10th triennial meeting, Washington, DC, USA, 22–27 August 1993. Burgio L., Clark R.J.H. & Firth S. 2001. Raman spectroscopy as a means for the identification of plattnerite (PbO2), of lead pigments and of their degradation products. Analyst 126: 222–227.

132

LACONA8.indb 132

11/10/2010 5:37:02 PM

Daniilia S., Sotiropoulou S., Bikiaris D., Salpistis C., Karagiannis G., Chryssoulakis Y., Price B.E. & Carlson J.H. 2000. Panselino’s Byzantine wall paintings in the Protaton Church, Mount Athos, Greece: a technical examination. Journal of Cultural Heritage 1: 91–110. Daniilia S., Minopoulou E., Demosthenous D. & Karagiannis G. 2008. A comparative study of wall paintings at the Cypriot monastery of Christ Antiphonitis: one artist or two? Journal of Archaeological Science 35: 1695–1707. Daniilia S. & Minopoulou E. 2009. A study of smalt and red lead discolouration in Antiphonitis wall paintings in Cyprus. Applied Physics A 96: 701–711.

Gavrichev K., Bolshakov A., Kondakov D., Khoroshilov A. & Denisov S. 2008. Thermal transformations of lead oxides. Journal of Thermal Analysis and Calorimetry 92 (3): 857–863. Kotulanová E., Bezdicka P., Hradil D., Hradilová J., Švarcová S. & Grygar T. 2009. Degradation of leadbased pigments by salt solutions. Journal of Cultural Heritage 10(3): 367–375. Piqué F. 1993. Scientific examination of the sculptural polychromy of Cave 6, Yungang, China. In N. Agnew (ed.), Conservation of ancient sites on the Silk Road, Proc. intern. Conf., Los Angeles, 3–8 October 1993.

133

LACONA8.indb 133

11/10/2010 5:37:02 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Comparative study on the irradiation methods against fungal colonization case study S.A. Abd Abd El Rahim Department of Conservation and Restoration, Faculty of Archaeology, Fayoum University, Fayoum, Egypt

ABSTRACT: Shaykhu Khanqah is one of two buildings face each other across Saliba street, Cairo, Egypt. It was built in (756 H–1355 A.D). Samples were collected from the deteriorated parts of the ground layer on the wood ceiling of Shaykhu Khanqah. These samples were analyzed to determine the fungal deterioration aspects. Seventeen different fungal species were isolated from the tested samples, belonging to the genera of Acremonium, Alternaria, Aspergillus, Aureobasidium, Cladosporium, Curvularia, Fusarium, Geotrichum, Mucor, Penicillium, Phoma, Rhizopus, Scopulariopsis, Stemphylium and Trichoderima. The data reveal that Cladosporium cladosporioides contributed the broadest spectra in the tested ground painting layer samples of the ceiling of Shaykhu Khanqah. Comparative sensitivity to radiation against all isolated fungal species indicate that the treatment of the tested fungal species with diode laser lead to complete inhibition of all tested species after 15 min exposure time. 1

INTRODUCTION

Shaykhu Khanqah is one of two buildings face each other across Saliba street, Cairo, Egypt. It was built in (756 H.–1355 A.D). This Khanqah included his mausoleum on the south of the street. It has a splendid painted wood ceiling. It has many deterioration aspects specially that due to the fungal deterioration (Williams, 1993). Fungi produce acids that decompose the monument material by producing salts and chelates that cause formation of cracks which may concentrate on the surface as crusts (Caneva, et al., 1991). It has to be recognized that fungi are involved directly and/or indirectly in the weathering of monuments and constituent minerals (Gomez, and De la Torre 1993). Conservators should consider biodeterioration processes as part of a complete and careful diagnosis of the decay in cultural objects. It is therefore very important to determine fungal genera growing on such mineral substrates and study their biology in order to understand the mechanisms of the adaptation processes. Wood ceiling is a cellulosic material, a preparation layer and paint layer upon it, may be easily degraded by microorganisms, as may the materials (glues) used to prepare a ground layer thus, besides the organic nature of the support, it contain organic molecules that many microorganisms may utilize for growth, such as sugars, gums, polysaccharides, proteins and oils (Strzelczyk, 1981). In painted works of art, the fungal deterioration processes can involve either a portion of the painting layer or all of its components (Strzelczyk et al., 1987). Wide

variations in the species isolated were reported in the analyses of the fungal flora present on painting layer (Agrawal et al., 1988). The fungal flora attacking the painting works include virtually all species of fungi because the variety of organic components of these works of art can represent a carbon source for practically all species, in addition, they show a great tolerance for environmental conditions and can use condensation moisture (Dhawan and Agrawal, 1986). The development of fungi on the surface of paintings induces aesthetical, mechanical and biochemical decay, where the growing mycelium spread over the paints masking design and colour, while the growth of hyphae and fruiting bodies inside the support can cause friability and loss of the paint layer (Tiano, 2001; Ionita, 1973) isolated 26 different species of fungi from stains appearing on the frescoes from areas of efflorescence and from zones in which the painted layer was fissured and portions were breaking away from the support. The species that are just saprophytes living on the painted surface may be growing at the expense of other microorganisms colonizing the frescoes, however, the idea that fungi may be the primary microbiological agents responsible for degradation of art works is so entrenched (Guglielminetti et al., 1994). Fungi such as Phanerochaete chrysosporium or Trameies versicolor are effective at degrading and discoloring dye due to the production of enzymes (Jarosz-Wilkolazka et al., 2002). Effect of cultural conditions on cellulases from Chaetomium indicum and Stachybotrys atra isolated from monumental objects was recorded

135

LACONA8.indb 135

11/10/2010 5:37:03 PM

by Darwish et al. (2005). The development of new technologies and materials to prevent fungal painting damage permits planning of restoration and conservation (Baldini, 1996). In order to control biodeterioration processes the control methods can be classified as mechanicals, physicals and chemicals. Traditional mechanical methods involve the physical removal of biodeteriogens either by hand or with tools, while mechanical methods can damage the substrate (Ionita, 1971). Chemical solvents have negative effects on the objects to be treated and not always effective against survival resistant or quiescent phase structures (Tiano, 2001). Murals restored were cleaned and treated with nystatin but showed the appearance of greenish brown to black spots on the painted surface (Sampo and Mosca, 1989). Several preventive and remedial methods have been used in tropical environments for control and eradication of microorganisms on the monuments. Remedial methods are aimed at the direct elimination by chemical treatments to eliminate and control the growth of biodeteriogens (Golubic et al., 2004). Physical methods include different types of radiation (gamma, UV and laser radiation). Gamma rays are a form of electromagnetic radiation used for sterilizing micro flora especially on organic materials such as paint (Van Der Molen et al., 1980). UV Solar radiation has a lethal effect on natural populations of culturable outdoor atmospheric microorganisms (Tong and Lighthart, 1996). UV irradiation used to disinfect indoor environments public shelters (Miller and Macher, 2000). There have been some reports about the effect of laser irradiation on microorganisms (Iwase et al., 1989). Several research groups confirmed the efficiency of ionizing radiation on biodeteriorating organisms (Magaudda, 2004). In order to evolve a suitable conservation programme, it becomes imperative to identify various organisms growing on ground painting layer on the wood ceiling as well as find the most suitable preventative maintenance non-destructive techniques employed on it. The final goal of this research is to characterize a series of markers to be used for the analysis of deterioration aspects and diagnosis of incipient pathologies to the knowledge of fungal mycoflora that occurs on ground painting layer on the wood ceiling in order to propose the most appropriate preventative maintenance non-destructive techniques employed on it.

cooled to just above the solidification and added to each Petri-dish. Swabbing with sterile cotton swabs and scalpel from markedly damaged surfaces of the tested ground layer with visible colonies of microscopic fungi was carried out. In the laboratory, swab samples were shaken mechanically for 10 min in 10 mL sterile distilled water and 1 mL aliquots of the resulting suspensions used to prepare spread plates on Czapeck’s Dox agar. Plates were incubated in the dark at 27°C for 7 days and the microscopic fungi were identified using the diagnostic keys of Gilman (l957), Barnett and Hunter (1972) and Moubasher (l993). The effect of different radiation on the isolated fungi:

2

3

MATERIALS AND METHODS

Isolation and identification of fungal colonization: Czapek–Dox’s agar media was used as isolation medium. The medium was brought up to 1000 mL with distilled water. Fifteen ml of this medium were

UV radiation: Spore suspension of the isolated fungal species were prepared in saline solution (0.85%, w/v, NaCl containing a drop of Tween 80) from 7 days old slant and irradiated with Phillips TUV -30-W-245 nm Lamp, type No. 57413-P/40 at a distance of 20 cm at exposure times (5, 10 and 15 min). The treated spores were kept in dark for 2 h. to avoid photoactivation repair. Gamma irradiation: The source of irradiation used for the tested fungal species was Cobalt-60 gamma cell 3500. This source is located at a Middle Eastern Regional Radioisotopes Center for the Arab countries (Dokki, Cairo). The dose rate was 2.4 Gy min-1. Fungal species were irradiated with dose of 25 kGy at exposure times (6, 12 and 18 h). Laser irradiation: Laser source was located at the National Institute of Laser Enhanced Science (NILES), Cairo University. The laser used was solid state diod laser of wave length 650 nm type DC Brushless PAT. Pending Crop. Tokyo. Japan. It was applied at output level of 250 mW. The total energy delivered was 30 J at exposure times (5, 10 and 15 min). All assays were conducted in triplicates. The irradiated spores were spread into plates containing Czapek–Dox’s agar media for five days incubation period. The growing colonies were counted against the control plates. The changes in survival caused by exposure to irradiation were evaluated by determining the relative recovery of microorganisms. The relative recovery was estimated as a ratio of number of colonies obtained with the impinge to the total number of microorganisms. RESULTS AND DISCUSSION

Source of isolation: In the present work the ground layer of the wood ceiling of Shaykhu Khanqah that belonged to the Mamluk period was built in (756 H.–1355 A.D). It located on Saliba street,

136

LACONA8.indb 136

11/10/2010 5:37:03 PM

Cairo, Egypt. A scientific study of the grounds was applied to know the components of this ground and also to study the types of fungal deterioration. The ground layer is the layer between the wood support and the paint layer. Fungal species that were isolated and its Frequency of occurrence: Fungi were isolated from different damaged areas of the tested ground layer of the ceiling of Shaykhu Khanqah in Table (1). The extent of fungal growth was assessed visually. Seventeen different fungal species were isolated belonging to the genera Acremonium, Alternaria, Aspergillus, Aureobasidium, Cladosporium, Curvularia, Fusarium, Geotrichum, Mucor, Penicillium, Phoma, Rhizopus, Scopulariopsis, Stemphylium and Trichoderima. The result in (Table 1) shows that the genus Cladosporium cladosporioides contributed the broadest spectra. The obtained data is in accordance with that obtained byNugari et al.

Table 1. Surveys of species and frequency of occurrence of fungi isolated from different sites on tested ceiling (colony/gm dry material).

11

M

1 and 3

13 15 5 13

M M L M

2 and 3 1, 2, 3 and 4 3 2 and 3

7

H

1, 2, 3 and 4

30

H

1, 2, 3 and 4

17

M

1 and 3

15

M

1 and 3

10 13 12 9 11

L M M L M

3 3 and 4 1, 2, 3 and 4 1 1, 2, 3 and 4

14

M

2 and 4

21

L

2

25 268 17

M – –

1 and 2 – –

Ultaviolet radiation Control

Ultaviolet radiation 5 min

Ultaviolet radiation 10 min

Ultaviolet radiation 15 min

100 90 80 70 60 50 40 30 20 10 tu m str

al te

ic

* High (H) 20–30, Moderate (M) 10–20, Low (L) 0–10.

rn at A. a ni ge A. r te rr A. eu s ve rs ic C. A. p olo r u cl ad llul a os po ns rio id C. es ge F. ni ch cu la la m ta yd os po G . c rum an d M idum .h ie m al P. gl is ab ru P. m he rb ar u R. m S. oryz br ae ev ic au S. lis ve sic T. aium ha m at um

0

A.

Acremonium strictum Alternaria alternata Aspergillus niger A. terreus A. versicolor Aureobasidium pullulans Cladosporium cladosporioides Curvularia geniculata Fusarium chlamydosporum Geotrichum candidum Mucor hiemalis Penicillium glabrum Phoma herbarum Rhizopus oryzae Scopulariopsis brevicaulis Stemphylium vesicaium Trichoderima hamatum Total count Number of species

Count of Freq. of Painting species occurrence* sites

A.

Fungal species

(1993) who reported that Cladosporium one of the major biological agents and the most significant agent responsible for fresco degradation and De la Torre et al. (1991) stated that the genera which have been demonstrated being more abundant on the monuments investigated are: Cladosporium, Penicillium, Trichoderma, Fusarium and Phoma and also dark spots are attributed to the presence of fungi of the family of Dematiaceae which contains, water, organic solvents insoluble and melanin pigments inside the mycelium. Bacterial colonization of the painted surface had chemically modified some of the components of the paint, rendering them utilizable by the fungus (O’Neill, 1986). In the opinion of the investigators, Cladosporium is one of the most commonly isolated flora from frescoes because it is resistant to variations in external factors (temperature, humidity) (Agrawal et al., 1989). Fungal deterioration of painted layer was discussed together with methods for their control by Sarbhoy et al. (1990). Aspergillus and Penicillium sp. were the most efficient in biodeterioration of restored frescoes (Sampo et al., 1990). Among the species of fungi most frequently involved in deterioration of the paint layer species of Penicillium, Aspergillus, Geotrichum develop on casein binders, Mucor and Rhizopus attack glue (Menier, 1988). Most of the common dominant tested fungal species are asexual this finding is in agreement with Karen (1994) who stated that Aspergillus, cladosporium, pullularia and penicillium were asexual fungi in all Cultural Historic materials. The effect of different radiation on the isolated fungi: The survival percentages of all tested species decreased gradually by increasing exposure time of UV- radiation till reached to the minimum value at 15 min. The most resistant pigmented species was C. cladosporioides with significant survival percentage of 14.4% (Fig. 1a). These results agree with that obtained by (Boyd-Wilson et al., 1998) who stated that the effects of short wavelength UV- radiation on spores diminished as pigmentation increased.

Figure 1a. UV-irradiation exposure times (LSD at 0.05 = 4.27).

137

LACONA8.indb 137

11/10/2010 5:37:03 PM

Gamma radiation Control

Gamma radiation 6 h

Gamma radiation 12 h

Gamma radiation 18 h

m

rn a

ni ge r te rr eu ve s rs ic ol A. or pu C. llu cl la ad ns os po r C. ioid es F. ge ch ni cu la m yd lata os po ru G m .c an di du m M .h ie m al P. is gl ab ru P. he m rb ar um R. or yz S. ae br ev ic au S. lis ve sic T. aium ha m at um A.

A.

A.

ic tu

te

str

al A.

A.

ta

100 90 80 70 60 50 40 30 20 10 0

Figure 1b. Gamma irradiation exposure times (LSD at 0.05 = 2.95).

Laser radiation Control

Laser radiation 5 min

Laser radiation 10 min

Laser radiation 15 min

100 90 80 70 60 50 40 30 20 10

um ar

yz ae br ev ic au lis S. ve sic ai um T. ha m at um

R.

or

S.

is

um br la

er b

P. h

m

ie m al

P. g

um or

di

du

M .h

sp

am yd o

G .c an

es

a

id

at

io

cu l

or

ge ni

sp do

C.

F. ch l

or

ns la

ic ol

llu

ve rs

pu

cl a

A.

C.

A.

a

ni ge r te rr eu s

A.

at

ic tu

te rn

str

al

A.

A.

m

0

A.

Distributions of fungi were significantly lower when UV lamps were used (Levetin et al., 2001). Caesar and Pearson (1983) found that UVradiation diminished the survival of ascospores of Sclerotinia sclerotiorum. The effect of germicidal effect of UV light on fungal flora was found by Menzies et al. (1999). UV- radiation is a major factor in Alternaria solani mortality (Rotem et al., 1985). Evaluation of the UV- irradiation induced changes in survivability of the tested fungal species suggests that some fungal species had a protection mechanism and have possibility to repair against UV- radiation. Ulevihus’ et al. (1999) stated that the high recovery potential of A. niger propagules and their ability to repair against UV- radiation with the time. The data in (Fig. 1b) reveal that the growth parameter of the tested fungal species decreased with increasing the exposure time of gamma radiation. The radioresistant fungal strains were C. cladosporioides, A. pullulans and A. alternata exhibited survival percentages 12.6, 10 and 7.6%, respectively after 18 h exposure time. Physical control (ultraviolet rays, gamma rays) have been used especially against fungi in the treatment of archaeological objects (De Cleene, 1994). The basic properties of paper are not significantly modified at gamma doses up to 10 kGy (Horakova and Martinek 1984). The treatment of the tested fungal species with diode laser lead to complete inhibition of all tested fungal species after 15 min exposure time (Fig. 1c). Difference in sensitivity to laser radiation at different exposure times may be due to the difference in cellular components. El-adly (1997) stated that the melanin pigment of some fungal species increases the absorbance of laser radiation and contributes in laser capture. Some microorganisms sensitized to killing by low power laser light (Wilson et al., 1993). Exposing the yeast cells to low power laser light rendering them susceptible to death (Dougherty et al., 1978). The concept of using a laser to treat painted surfaces is to avoid the use and the disadvantages of solvents (Swicklik, 1993). Diode laser

Figure 1c. Effect of diode laser- irradiation exposure times (LSD at 0.05 = 6.11) on the survival of isolated fungal species.

irradiation was the most efficient radiation type for all tested species and C. cladosporioides was the most radioresistant species. 4

CONCLUSIONS

An alternative is the use of laser radiation, a promising treatment in the preservation field. The obtained results confirmed that diode laser radiation treatment is extremely efficient. The laser preservation technology has brought a powerful way to save our culture heritage from being damaged by molds, guaranteeing a good quality of life for the employees and users.

ACKNOWLEDGEMENT The author thanks Dr. N.S. Gweely in the Faculty of Science, Botany Department, Cairo University, Giza, Egypt for her assistance and helpful discussions.

REFERENCES Agrawal, O.P., S. Dhawan, F. Garg, N. Shaheen and A. Misra. 1988. Study of biodeterioration of the Ajanta wall paintings. Int. Biodeterior.24:121–129. Agrawal, O.P., S. Dhawan and K.L. Garg. 1989. Microbial deterioration of paintings a review. Intach Conservation Centre, Lucknow, India. 1–51. Baldini, U. 1996. Teoria del restauro e unità di metodologia, 1–2, Nardini, Firenze. Barnett, H.L. and B.B. Hunter. 1972. Illustrated of imperfect fungi, 3rd Ed., Burgess Publishing Co., Minneopolis, Minnescoata. Boyd-Wilson, K.S.H., J.H. Perry and M. Walter. 1998. Persistence and survival of saprophytic fungi antagonistic to Botrytis cinerea on Kiwifrutt leaves. Proc. 51st N.Z. Plant Protection Conf., 96–101. Caneva, G., M.P. Nugari, and O. Salvadori, 1991 Biology in the conservation of the works of art. Rome:ICCROM. Darwish, S.S., M.A. Abdou, and M.A. Maysa, 2005. Effect of cultural conditions on production of cellulases from

138

LACONA8.indb 138

11/10/2010 5:37:03 PM

Chaetomium indicum and Stachybotrys atra isolated from monumental wooden objects. Conference and workshop on conservation and Restoration, Faculty of fine arts, Minia Univ. De Cleene, M. 1994. Summary of the results, General conclusion and Recommendation. The Role of Heterotrophic Bacteria in the Degradation of Stone. In Environment Protection and Conservation of the European Cultural Heritage. Research Report n°2. M De Cleene Ed., Science Information Office University of Ghent Publ., 261. De la Torre, M.A., G. Gomez-Alarcon, P. Melgarejo and C. Saiz-Jimenez. 1991. Fungi in weathered sandstone from salamanca cathedral, Spain. The Sci. Tot. Environ., 107: 159–168. Dougherty, T.J., J.E. Kauiman, A. Goldfarb, K.R. Weishaupt, D. Boyle. and A. Mittleman. 1978. Photoradiation therapy for the treatment of malignant tumours. Cancer Res., 38: 2628–35. El-Adly, A.A. 1997. Studies on the effect of laser on growth and metabolism of some microorganisms. M.Sc. Thesis, Ain Shams University. Gadd, G.M. and A.J. Griffiths, 1980. Influenceof pH on toxicity and uptake of copper in Aurebasidium pollulans. Trans. British. Mycol. Soc., 75: 303–317. Gilman, J.C. 1957. A manual of soil fungi. The lowastate college press. Ames, Iowa, USA pp: 450. Golubic, S., R. Gudrun, and L. Threrese, (2004): Endolithic fungi in marine ecosystems. Trends in Microbiology. 13: 229–235. Gomez, A. and M.A. De la Torre, 1993. The effect of filamentous fungi on stone monuments. Building mycology, London (In Press). Horakova, H. and F. Martinek. 1984. Disinfection of archive documents by ionizing radiation. Restaurator., 6: 205–216. Ionita, I. 1971. Contribution to the study of the biodeterioration of the works of art and of historic monuments. 11. Species of fungi isolated from oil and tempera paintings. Rev. Rourn. Biol. Botanique., 16: 377–381. Ionita, I. 1973. Contributions to the study of the biodeterioration of the works of art and historical monuments. IV. Fungi involved in the deterioration of mural paintings from the monasteries of Moldavia. Rev. Roum. Biol. Ser. Bot. 18: 179–189. Iwase, T., N. Hori and T. Morioha. 1989. Possible mechanism of He-Ne laser effects on the cell membrane characteristic. Laser Med. Surg. 4: 166–171. Jarosz-Wilkolazka, A., J. Kochmanska-Rdest, E. Malarczyk, W. Wardas and A. Leonowicz. 2002. Fungi and Their Ability to Decolourize Azo and Anthraquinonie Dyes, Enzyme Microb. Technol. 30: 566–572. Karen, M. 1994. Non-Toxic Fumigation and Alternative Control Techniques for Preserving Cultural Historic Properties and Collections Notes on a Conference. LeDuy, A. 1986. Cellulase from an acidophilic fungus. Biotechnol. Renewable Energy. Moo-Young, M. Hasnain, S. Lamptey, J. eds. 93–100. Levetin, N.E., R. Shaughnessy, A. Christine and R. Scheir. 2001. Effectiveness of Germicidal UV Radiation for Reducing Fungal Contamination within Air-Handling Units. App. Environ. Microbiol., 67: 3712–3715.

Magaudda, G. 2004. The recovery of biodeteriorated books and archive documents through gamma radiation: some considerations on the results achieved. J. Cult. Herit., 5: 113–118. Menier, J. 1988. Sur quelques insectes deprédateur des archives, Patrimoine culturel et al terations biologiques. Actes des journees d’etudes de la S.F.I.I.C., Poitiers, 17–18 November. 45–52. Menzies, D.J. Pasztor, T. Rand. and J. Bourbeau. 1999. Germicidal ultra-violet irradiation in air conditioning systems: effect on office worker health and wellbeing: a pilot study. Occup. Environ. Med. 56: 397–402. Miller, S.L. and J.M. Macher. 2000. Evaluation of a methodology for quantifying the effect of room air ultraviolet germicidal irradiation on air- borne bacteria. Aerosol Sci. Technol. 33:274–295. Moubasher, A.H. 1993. Soil Fungi in Qatar and other Arab Countries. Sci. App. Res. Center, Univ. Qatar. p. 566. Nugari, M.P., M. Realini and A. Roccardi. 1993. Contamination of mural paintings by indoor airborne fungal spores. Aerobiologia 9: 131–139. O’Neill, T.B. 1986. Succession and interrelationships of microorganisms on painted surfaces. J. Coatings Technol., 58: 51–56. Ross, I.S. 1982. Effect of copper, cadmium and zinc on mycelial growth of Candida albicans. Trans. British. Mycol. Soc., 78: 543–545. Rotem, J., B. Wooding and D.E. Aylor. 1985. The role of solar radiation especially ultraviolet in the mortality of fungal spores. Phytopathol. 75: 510–514. Sampo, S. and A. Mosca. 1989. A study of the fungi occurring on 15th century frescoes in Florence, Italy. Int. Biodeterior. 25: 343–353. Sampo, S., G. Pascal, A. Reisinger and A. Bresinsky. 1990. Efficiency of fungi in biodeterioration of restored frescoes: an evaluation by growth curves on materials used in restoration. Abstracts of the Fourth International Mycological Congress, IMC 4, held at Regensburg, Germany, 28 Aug.–3 Sep. 252–3. Sarbhoy, A.K., S.K. Hasija and S.H. Bilgrami. 1990. Role of fungi in biodeterioration of stones, leather, fuel, paintings, rubber and gunpowder. Persp. Mycolog. Res., 2: 219–225. Somers, E. 1963. The uptake of copper by fungal cells. Ann. Appl. Biol., 51: 425–437. Strzelczyk, A.B. 1981. Paintings and sculptures. In A.H. Rose (ed.), Microbial deterioration. Academic Press, London, United Kingdom. 203–234. Strzelczyk, A.B, J. Kuroczkin and W.E. Krumbein, 1987. Studies on microbial degradation of ancient leather bookbindings: part l. Int. Biodet. Bull., 23: 3–27. Swicklik, M. 1993. ‘French painting and the use of varnish, 1750–1900’ In Conservation Research, National Gallery of Art, Washington DC, 157–174. Tiano, P. 2001. Biodegradation of Cultural Heritage: Decay Mechanisms and Control Methods. CNR Centro di studio sulle “Cause Deperimento e Metodi Conservazione Opere d’Arte”, Via G. Capponi 9, 50121 Firenze, Italy. Tong, Y. and B. Lighthart. 1996. Solar radiation has a lethal effect on natural populations of culturable outdoor atmospheric bacteria. Atmos. Environ., 31: 897–900.

139

LACONA8.indb 139

11/10/2010 5:37:03 PM

Ulevihus’,V., D. PeEiulyte, A. Juozaitis’ and A. Lugauskas. 1999. Survival of airborne fungal propagules exposed to ultraviolet irradiation. J. Aerosol Sci., 30: S815–S816. Van Der Molen, L, J. Garty, B.W. Aardema and W. Krumbein. 1980. Growth control of algae and cyanobacteria on historical monuments by a mobile UV unit (MUVU), Studies in Cons., 25: 71–77.

Wilson, M., J. Dohson and S., Sakkar 1993. Sensitization of periodontopathogenic bacteria to killing by light from a low power laser. Oral Microbial Immunol, 8: 182–187. Willams, C., 1993. Islamic monuments in Cairo, A practical guide, 4th Ed. The American University in Cairo Press, 61.

140

LACONA8.indb 140

11/10/2010 5:37:03 PM

Investigation and diagnostics methods

LACONA8.indb 141

11/10/2010 5:37:03 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Absolute LIBS stratigraphy with Optical Coherence Tomography P. Targowski, E.A. Kwiatkowska & M. Sylwestrzak Institute of Physics, Nicolaus Copernicus University, Toruń, Poland

J. Marczak, W. Skrzeczanowski & R. Ostrowski Institute of Optoelectronics, Military University of Technology, Warszaw, Poland

E. Szmit-Naud & M. Iwanicka Institute for the Study, Restoration and Conservation of Cultural Heritage, Nicolaus Copernicus University, Toruń, Poland

ABSTRACT: In this contribution the application of Optical Coherence Tomography (OCT) as a complementary tool for Laser Induced Breakdown Spectroscopy (LIBS) depth profile or stratigraphy analysis. It is shown how OCT may be used for preliminary evaluation of locations of perspective LIBS examination, as an aid for proper adjusting the system and mostly for absolute in-depth calibration of LIBS data. Additionally, the inspection of OCT images of transparent part of paint layer combined with the analysis of the ablation rates during the LIBS depth profile analysis enables one to distinguish certain strata in the sample in order to aid in the analysis of LIBS data. The application of this approach to revealing stratigraphy of a historic easel painting is presented and discussed. 1

INTRODUCTION

Laser Induced Breakdown Spectroscopy (LIBS) (Fotakis et al. 2007) is a well established analytical technique used for determination of the elemental composition of materials. This is achieved by spectral examination of atomic specific emission generated in a plasma plume formed by a high power laser pulse focused on the object surface. Since the energy is focused on a small spot and resultant ablation crater has a diameter less then a few hundred micrometers, the LIBS technique may be considered a micro-invasive. After about a microsecond from the laser blast, a broadband emission from free electrons decays and sharp lines of characteristic emission of elements embedded in the examined sample are visible and available for analysis. In such a way a global information of elemental composition is retrieved. In order to resolve the composition of successive strata of the object under examination, consecutive laser pulses are applied to the same place. Their energy is adjusted at the low level to ensure ablation only of a thin layer of the material in one pulse. This technique is called LIBS profile analysis or LIBS stratigraphy (Anderson et al. 1995, Mendes et al. 2009). The disadvantage of this approach lays in lack of absolute information on the depth from which certain spectrum is collected. In result the determination of layer thickness is not possible. Moreover, since the ablation rate varies for different strata

within the paint layer, obtained profile becomes distorted nonlinearly, if presented as a function of the laser pulse number. To overcome this drawback the Optical Coherence Tomography (OCT) has been used lately (Amaral et al. 2009, Kwiatkowska et al. 2009). In this contribution the preliminary results on innovative use of this technique for monitoring the LIBS study of easel painting are presented for the first time. OCT is a technique for non-invasive imaging of cross-sections of semi-transparent objects with near IR light (Stifter 2007, Drexler et al. 2008, Targowski et al. 2008, for complete list of papers see also www.oc4art.eu) In this specific case OCT can be used firstly as an aiming tool aiding proper selection of the place of testing with LIBS and then for monitoring the ablation process. Specifically, in case of LIBS stratigraphy, the depth of ablation crater may be directly measured after each laser pulse and thus the stratigraphic profile may scaled in depth absolutely. 2

INSTRUMENTATION AND METHODOLOGY

2.1 The LIBS system In this study the portable LIBS system developed and operated in the Institute of Optoelectronics, Military University of Technology, Warsaw, Poland was used. The equipment was transferred

143

LACONA8.indb 143

11/10/2010 5:37:04 PM

to Torun, to avoid transportation of the painting and assembled in the Institute of Physics, NCU to integrate with the OCT tomograph. To create plasma, gentle pulses of fourth harmonic of Q-Switched Nd-YAG laser (Brio from Quantel, France, λ = 266 nm, pulse energy = 12 mJ at object, pulse duration ∼4 ns) were used. Echelle Spectra Analyzer ESA 4000 from LLA Instruments GmbH was used to analyze radiation from plasma. The instrument was equipped with a MCP (gated microchannel plate) image intensifier and an iCCD sensor array (Kodak KAF-1001). The spectral range of the Echelle spectrometer was 200 nm to 780 nm with a linear dispersion varying from 5 pm/pixel at 200 nm to 19 pm/pixel at 780 nm. Theoretical spectral resolution λ/Δλ was about 20,000. Spectral data were collected for 10 μs with 1 μs delay after laser pulse to avoid background radiation from free electrons. To analyze spectra ESAWIN 4.1.4 software with spectral data base ROI Service were used. 2.2

The OCT tomograph

After each pulse the volume OCT data were collected with a prototype tomograph build in the Institute of Physics, NCU especially for application in conservation science. This high resolution instrument is described in details elsewhere in this volume (Iwanicka et al. 2009). Briefly: a narrow beam of probing, infrared radiation is scanned over the object, scattered at the surface as well as at elements of object’s structure, and finally examined by means of interference. The instrument is thus capable of creating cross-sectional images (tomograms, Figs. 2, 3) of objects weakly absorbing light. In OCT axial and transversal resolutions are decoupled. In the instrument described the former is equal to 4 μm (in air) when the latter is about 30 μm. Together with the instrumental limit of in-depth imaging of about 2 mm, all this make OCT tomography especially suitable for imaging layered structures like paintings, for instance. Tomograms are shown in gray scale: high scattering centres are shown as dark spots and low scattering media remains white or light gray. For instance, in Figure 2, the uppermost visible line is the air-varnish interface. Then 3 layers of semi-transparent media are present. Below, the opaque paint layer limits the depth available for OCT examination. Cross-sectional data collected in parallel, adjacent positions may be also combined into volume 3D data. From such data the surface elevation map may be generated automatically. OCT data were used for determination of ablation crater depth as described previously (Kwiatkowska et al. 2009) and for determination of the volume of abated material. For the later, two elevation maps were directly used: one obtained

before ablation and second collected after the given pulse. The volume of the crater was thus calculated as a sum of differences of elevations at every point of the surface multiplied by the voxel volume. 2.3 Experiment set-up For this study the experiment set-up was assembled in a simple manner (Fig. 1). To avoid geometrical distortions of profilometric data, the OCT tomograph’s scanning beam was set perpendicularly to the object’s surface. Since no dichroic mirror were used, the LIBS ablation pulses were delivered with a focusing lens at angle of incidence of about 45°. The fluorescence were collected at angle of about 60°. In future it is planned to equip the system with the dichroic mirror to ensure the perpendicular incidence of ablation pulses. The first instrumental advantage of using OCT together with the LIBS set-up is that it assists in adjusting the system. For achieving credible results it is necessary to have the fluorescence collecting optics precisely focused at the plume. To do so it is essential not only to have focal points of ablating beam and fluorescence collecting systems co-focused, but also the target surface must be placed precisely in this point. Whilst the former can

Figure 1.

Experiment arrangement—the top view.

Figure 2. OCT cross-section taken before the LIBS ablation was performed.

144

LACONA8.indb 144

11/10/2010 5:37:04 PM

be done once and is simple to achieve by targeting at a test sample, adjusting with the real target must be repeated for every point of examination. This can be easily done with aid of a real time OCT imaging if the OCT object scanner is fixed with the LIBS system. Thus the targeting is performed by observing the OCT preview and placing the object surface at proper depth with a micrometer precision. Additionally, place for the further analysis may be checked for a stratigraphy of upper layers and sites of previous conservations/alterations may be excluded. The experiment was performed as follows: firstly the sub-surface structure of the site was checked with OCT (Fig. 2) to avoid previous local alterations. After that the OCT volume data (400 parallel slices) were collected as a reference. In the next step laser pulse was applied and LIBS data collected. Subsequently the site was scanned with OCT again to collect the next set of volume data. Then the sequence of LIBS data collection followed by OCT scanning was repeated until the prime layer of the picture was reached. In Figure 3

Figure 3.

OCT cross-section after four LIBS pulses.

Figure 4. Virgin and Child—oil on canvas. Dots mark places where LIBS analysis were performed. Results obtained at spot #1 are discussed in this contribution.

an exemplary OCT cross-section of the ablation crater (after 4 pulses) is shown. 2.4 About the object In this contribution application of our method to the real object is reported for the first time. For this purpose an oil painting on canvas of unknown time and place of origin depicting the Virgin and Child was chosen (Fig. 4). The picture is well preserved and thus the possibility of sampling is strictly limited. The problem to be resolved was the authenticity of certain parts of a paint layer, suspected to have been painted over old and deteriorated one. 3

RESULTS

3.1 Ablation rates determination In this paragraph results of combine LIBS and OCT analysis performed at spot #1 (Virgins veil) will be presented. The ablation crater’s depths determined from OCT data are shown in Figure 5 as the function of the pulse number. It is clear from the figure, that the slope of this relation, namely the ablation speed, vary within the experiment. In the same graph corresponding crater volumes are shown. It is evident that, despite of minor differences, that these two relationships coincide until 21st pulse. After that, the increase of the crater volume is no longer accompanied with the rise of its depth. Inspection of surface maps leads to a conclusion, that only the diameter of the crater increases at this stage of experiment. Therefore, it may be assumed that during these pulses the material is ablated from the walls of the crater only and hence is not specific for any particular depth. Accordingly, data from pulses No. 22 were not analysed for the elementary contents. Moreover, by close inspection of OCT crosssections (Figs. 2, 3) it is possible to identify the layers within the part of paint stratum transparent to

Figure 5. Virgin and Child—spot #1, Virgin’s veil; ablation crater depth and volume as function of laser pulse. Vertical lines show layer boundaries—see text for details.

145

LACONA8.indb 145

11/10/2010 5:37:04 PM

infrared OCT light. The boundaries of these layers (marked as I and II) lay at about 30 μm and 40 μm under the paint’s surface. Positions of these interfaces are marked in Fig. 5 with dash-dot lines. It is evident that these layers are reached in 3rd and 5th pulses respectively. It is worthwhile to note that while the ablation rate after the 3rd pulse remains the same, it changes drastically after the 5th one. Similar result was obtained previously with a test sample (Kwiatkowska et al. 2009). Since all binding media contained in paint layer absorb UV laser radiation very strongly, this change in ablation rates is associated rather with their chemical composition: similar for layers I and II and different in opaque (for IR) layers located deeper. Despite of the origin of this effect, possible utilisation of such data for stratigraphy is evident. 3.2 LIBS data analysis Analysis of spectral LIBS data was performed for pulses from No. 2 until 21. The reason of rejecting data from later pulses is given in the previous section. First pulse was discarded from the spectral analysis because it is considered as a “cleaning” one and it’s spectra contains all surface contaminations. Elements and their spectral lines investigated for this sample are summarised in Table 1. If possible, Table 1.

Elements analysed for spot #1.

Element Wavelength (nm) Comments Na Mg Al Si P K Ca Ti Cr Mn Fe Co Cu

589.591* 285.213 394.403* 288.157 253.565* 769.898* 445.476 453.324 428.971* 279.481 438.351 346.579 324.752

Zn As Se Sr Mo Cd Sn Sb Ba Pb

– – – 460.733* 390.295* 346.619 – 252.852* 455.403* 405.782

*

present in all layers present in all layers see Fig. 7 see Fig. 7 very week week, present in all layers similar to Cd very week week in I-III, strong in IV,V present from layer IV see Fig. 7 similar to Cd week, present in layers II-IV not present not present not present see Fig. 6 week, present in all layers see Fig. 6 not present very week very week present from layer II

Energy of excited state is out of range 3.5–5.5 eV.

for analysis lines with their excited states of similar energy of 4.5 eV were chosen. Otherwise, the strongest other lines were selected. For some elements only traces of their presence were found, and for others their presence can not be confirmed. However, they are listed in the Table because this result may be important for pigment identification purpose. Signals from others were not specific for any particular layer. Therefore five most interesting elements were chosen for presentation. In Figure 6 presence of three elements: Si, Al, and Fe is shown. Apparently, all these elements are not present in layer I and their concentrations (related to line intensities) increase at depth of about 30 μm—in coincidence with beginning of the layer II. Interestingly, there is no significant difference at depth of 40 μm—the concentrations in further strata are similar until the depth of 117 μm. From this position (equivalent to 14th pulse) all three concentrations increase twice. Therefore it is reasonable to identify a separate strata from this depth. It is marked in Fig. 5 with a number V. In Figure 7 the results of similar analysis performed for presence of strontium and cadmium are shown. In this case both line intensities increase significantly much later—at depth of 80 μm, with the 8th laser pulse. This gives an argument for localisation of the another boundary between layers (marked as III and IV). Additionally, the analysis of signal from strontium confirms again the presence of the layer V, located between 117 μm and 135 μm.

Figure 6. Relative intensities of silica, aluminium and iron lines at spot #1 as detected by LIBS.

Figure 7. Relative intensities of strontium and cadmium lines at spot #1 as detected by LIBS.

146

LACONA8.indb 146

11/10/2010 5:37:05 PM

Therefore, from Figures 6 and 7 it is evident, that below the layer III there are at least two more layers, of different composition. Since the layer III is completely opaque there seems to be no reason for such a complicated structure other than later alteration of primary composition, build of layers IV and V. The foregoing discussion based on both OCT and LIBS analysis may be summarised by suggesting the following composition of the paint layer at spot #1: – layer I: two strata of varnish, as seen from Fig. 2, – layer II: semi-transparent glaze over a secondary varnish, – layer III: secondary opaque yellow paint layer, – layer IV: original (?) paint layer of different composition in comparison to layers II and III, – layer V: —original (?) prime. Detailed determination of pigments embedded in these layers needs further investigation and is beyond the scope of this report, dedicated mainly to the discussion of the technique. 4

CONCLUSIONS

In this contribution, using an exemplary painting from conservation studio, it is demonstrated how OCT may complement LIBS analysis. Firstly, it assists in recognizing proper locations for examination. It is important to avoid areas of previous conservation treatments, places with very inhomogeneous structure like cracks, clods of paint etc. Secondly, the real time OCT preview aids in placing examined object at a proper distance, with the common focal point of LIBS laser and fluorescence collecting optics at object’s surface. Thirdly, in case of LIBS stratigraphy, collecting the OCT volume data after each laser pulse enables resolving of the depth of the present ablation crater. This permits an absolute in-depth calibration of LIBS elemental profiles even below the OCT imaging range (limited by the transparency of object’s strata). Additionally, it is often possible to differentiate subsequent layers of the material: either by direct observation of the structure at the OCT crosssectional image or by seeking changes in the ablation rates. Whereas the former is possible within a transparent layers of the object (varnishes and glazes), the latter may be performed anywhere within the profile. However, it must be emphasized that constant ablation rate does not imply the homogeneity of the structure. For the proper analysis of the structure it is thus essential to combine information both from OCT and LIBS data.

ACKNOWLEDGEMENTS This work was supported by Polish Government Research Grants through the years 2008–2011. MI, EK, and MS gratefully acknowledge additional support from the European Social Fund and Polish Government within Integrated Regional Development Operational Program, Action 2.6, by project “Stypendia dla doktorantów 2008/2009— ZPORR” of Kuyavian-Pomeranian Voivodeship. Additionally, MI acknowledges support from the project operated within the Foundation for Polish Science Ventures Programme, financed by the EU European Regional Development Fund.

REFERENCES Amaral, M.M., Raele, M.P., de Freitas, A.Z., Zahn, G.S., Samad, R.E., Vieira, Jr., N.D. & Tarelho, L.V.G. 2009. Laser Induced Breakdown Spectroscopy (LIBS) applied to stratigrafic elemental analysis and optical coherence tomography (OCT) to damage determination of cultural heritage Brazilian coins. Proceedings of SPIE 7391: 73910I. Anderson, D.R., McLeod, C.W., English, T. & Trevor Smith, A. 1995. Depth Profile Studies Using LaserInduced Plasma Emission Spectrometry. Applied Spectroscopy 49(6): 691–701. Drexler, W. & Fujimoto, J.G., (eds.) 2008. Optical Coherence Tomography: Technology And Applications. Biological And Medical Physics, Biomedical Engineering. Berlin Heidelberg New York, SpringerVerlag. Fotakis, C., Anglos, D., Zafiropulos, V., Georgiu, S. & Tornari, V. 2007. Lasers in the Preservation of Cultural Heritage. Boca Raton, CRC Press. Iwanicka, M., Tymińska - Widmer, L., Rouba, B.J., Kwiatkowska, E.A., Sylwestrzak, M. & Targowski, P. 2009. Through-glass structural examination of Hinterglasmalerei by Optical Coherence Tomography. In Lasers in the Conservation of Artworks, LACONA VIII Proceedings - in this volume. Kwiatkowska, E.A., Marczak, J., Ostrowski, R., Skrzeczanowski, W., Sylwestrzak, M., Iwanicka, M. & Targowski, P. 2009. Absolute LIBS stratigraphy with Optical Coherence Tomography. Proceedings of SPIE 7391: 739114. Mendes, N.F.C., Osticioli, I., Striova, J., Sansonetti, A., Becucci, M. & Castellucci, E. 2009. Versatile pulsed laser setup for depth profiling analysis of multilayered samples in the field of cultural heritage. Journal of Molecular Structure 924–926: 420–426. Stifter, D. 2007. Beyond biomedicine: a review of alternative applications and developments for optical coherence tomography. Applied Physics B Lasers and Optics 88(3): 337–357. Targowski, P. Rouba, B., Góra, M., Tymińska-Widmer, L., Marczak, J. & Kowalczyk, A. 2008. Optical coherence tomography in art diagnostic and restoration. Applied Physics A: Materials Science and Processing 92: 1–9.

147

LACONA8.indb 147

11/10/2010 5:37:05 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Database of complex paint spectra decomposed by principal component analysis, for identification of artwork colours Zs. Márton & T. Tóth Institute of Physics, University of Pécs, Pécs, Hungary

É. Galambos Museum of Fine Arts, Budapest, Hungary

R. Mingesz Department of Experimental Physics, University of Szeged, Szeged, Hungary

ABSTRACT: Analysis is a main issue in artwork restoration and protection. The method should be non-destructive—first of all—but also mobile, reliable, quick, cheap, safe and easy to handle. At the same time, the colours of an artwork consist of different layers of complex materials, often of unknown origin. It will be shown how Laser Induced Breakdown Spectroscopy (LIBS) combined with Principal Component Analysis (PCA) can contribute to paint identification in realistic cases. In order to get large amount of spectral information from the paint layer, a high-resolution, wide-range spectrograph, Andor Mechelle 5000, was used for detection of the LIBS spectra of the samples. Certain characteristic spectral lines and features, like high intensity lead or copper lines are easily recognisable at first sight, but they are not sufficient for characterisation of a complex paint layer. In this study, we make use of the capability of LIBS of sampling a small, but macroscopic area, thus showing the average elemental composition in a single measurement. Spectral database was collected from previously prepared paint samples of known origin. Blue and green pigments applied with different binding materials over the same white ground layer were examined. LabView programmes were prepared for performing PCA on the measured spectra. Influence of data base choice, of binding materials and their aging, as well as that of percentage composition of mixtures on the PCA results are evaluated in the present paper. Keywords: 1

pigment, laser induced breakdown spectroscopy, principal component analysis

INTRODUCTION

A large number of sophisticated elemental analysis techniques are used in artwork characterization (Ciliberto 2000). They provide either qualitative or quantitative information on the elemental composition, and differ in accuracy, in sensitivity, in spatial resolution, in requirements for sample preparation, in working principle, in destructivity, in safety and other parameters. Some of them, like X-Ray Fluorescence (XRF) and Particle Induced X-Ray Emission (PIXE) are sensitive down to the ppm level (Brundle 1992). In spite of its relative simplicity, Laser Induced Breakdown Spectroscopy (LIBS) can compete with those other techniques. Its applicability relies in its straightforwardness, quickness, minimal invasiveness, and in the fact that it can be applied in situ, without sample preparation. Portable LIBS systems are also present on the market. Spatial

resolution of LIBS can be increased to a few microns (Osticioli 2008). Depending on the spectrograph, a vast amount of spectral data can be collected in seconds. However, to gain quantitative results from a LIBS measurement is still a challenge, although there exist techniques for that (Cremers & Radziemski 2006). In the present work it is shown how the performance of LIBS can be increased by making use of the large amount of information contained in the high-resolution, wide-range spectra and of the reference database that can be easily collected. Instead of aiming the analysis of microscopic components of the material of complex paints, the robustness of the technique is utilised, and conclusions are drawn on the average composition of the paint layers. This is done with the aid of a well-known chemometric method, Principal Component Analysis (PCA).

149

LACONA8.indb 149

11/10/2010 5:37:05 PM

2

EXPERIMENTAL AND DATA PROCESSING METHODS

2.1

Table 3.

Sample preparation

A great advantage of LIBS is, that there is no need for any particular sample preparation, as long as the target of investigation is accessible for the exciting laser beam and for the optical fiber collecting the emitted light. Our samples were 15 × 15 mm2 painted spots, containing known pigments and different binding materials respectively. The codes starting with B, and G15 refer to freshly painted layers, while rest of the samples were prepared in back in the year 2000, originally for a different purpose. The chemical composition of the pigments was confirmed with X-Ray Diffraction (XRD), polarization microscopic and X-Ray Fluorescence analysis (XRF). The binding materials were commercially purchased. Their source, manufacturer, or trademark is indicated in Table 1. Table 2 specifies the painted samples referred to in the present study,

Pigment

Chemical composition

Cobalt blue Green earth

CoO . Al2O3 Complex, Mg, Al, Si, S, Cl, Ca, Ti, Cr, Fe 2CuCO3 . Cu(OH)2 Cu(CH3COO)2 ZnO, CoO Cr2O3 Cr2O3 . 2H2O

Malachite Copper acetate Rinmann’s green Chromium oxide Viridian

while the chemical composition of the referred pigments is given in Table 3 (Eastaugh 2005). In this paper, the word “paint” shall be used for colour layers containing pigment and binding material, applied over a ground layer consisting of gypsum and chalk, in order to make clear distinction between the complex paints and pure pigments. 2.2

Table 1.

Origin of the applied binding materials.

Binding

Source, trademark

Glue Gelatine Egg yolk Dextrin Gum Arabic Linseed oil Damar Plextol B500 Egg white Paraloid B72

Rabbit skin glue, Natural Pigments Edible gelatine, Dr. Oetker Local market Pannon Color 26021331 Kremer 63300 Art Export 4304176 Art Export 4304305 Lascaux Farbenfabrik Local market Lascaux Farbenfabrik

Table 2.

Composition of paint samples. Pigments

Composition

Code

1

2

B100 B96 B85 B50 B27 B0 G1_6 G3_6 G4_6 G7_6 G8_6 G8_8 G9_6 G15

Cobalt: Prussian blue Cobalt: Prussian blue Cobalt: Prussian blue Cobalt: Prussian blue Cobalt: Prussian blue Cobalt: Prussian blue Green earth Malachite Copper acetate Rinmann’s green Chromium oxide Chromium oxide Viridian Chromium oxide

[1:2] w/w% Binding 100 96 85 50 27 0

Linseed oil Linseed oil Linseed oil Linseed oil Linseed oil Linseed oil Glue Glue Glue Glue Glue Linseed oil Glue Glue

Composition of pigments.

LIBS

LIBS was carried out in standard arrangement (Fotakis 2006. p. 57). KrF excimer laser (248 nm, 16 ns) was used as an irradiation source. The laser pulses were imaged onto the sample surface by a fused silica lens, and both the sampling spot size (∼1.1 mm2) and irradiating fluence (2.8 J/cm2) were kept constant. By choosing such a large sampling area and high laser fluence, the available signal to noise ratio is high. Furthermore, this way, sampling immediately performs a space averaging over the examined layer, which is inherently inhomogeneous in most of the arts objects. This approach is the reverse of the one that Osticioli and his colleagues use (Osticioli 2008). Instead of focusing on one single grain of pigment, we collect information from the complex paint layer in one step. The light emitted from the ablation plasma plume was collected through an optical fiber into an Andor Mechelle 5000 spectrograph. An iStar DH734-18F-03 camera (Andor Technology)— equipped with digital delay generator—recorded the LIBS spectra. Typical delay and gate times were 500 ns and 600 ns, respectively, so that the continuous background could be eliminated, and the noise could be reduced. Overall spectral resolution of the system is 0.05 nm @500 nm, while the spectral width is large (250–800 nm), due to the echelle technology. It is important to note that the echelle grating efficiency drops to about half of its maximum at the extremities of each order (Schroeder & Hillard 1980). This effect was corrected by the so-called y-calibration with the aid of an Ocean Optics DH-2000 combined deuterium-halogen standard lamp.

150

LACONA8.indb 150

11/10/2010 5:37:05 PM

The spectra recorded this way are reasonably clear, repeatable and rich in details. Nevertheless, identification of spectral lines in them is still a challenge, since many factors influence the emission intensity in LIBS: the laser-to-sample coupling efficiency, quantity of neutral and ionized species within the plasma, collisional interactions and self-absorption (Huang 2004, Diwakar 2007). In the present study, mainly qualitative analysis of paints was aimed. One way of evaluation of the spectra is comparing them with standard spectra of pure materials. This method usually allows the identification of the most intensive characteristic elemental lines, but still does not account on the matrix effect. At the same time, it is tedious, considering the ∼20000 intensity data that a single spectrum consists of. For clarifying the patterns hidden in the abundance of data, it makes sense to turn to Principal Component Analysis (PCA). 2.3 PCA PCA is a standard tool of modern data analysis. It reduces the dimensionality of a large number of interrelated variables by algebraic calculations, with retaining as much of the information about the original data set as possible (Jolliffe 2002). Our spectra consist of intensity data corresponding to the pixels of the camera, which represent different wavelengths. Thus, a spectrum can be considered as a vector in the ∼20000 dimension space of wavelengths. PCA means that we take a set of such spectra (the so-called data matrix) and find a set of linearly independent variables (Principal Components, PCs) by means of which the original spectra can be expressed. Mathematically, PCs are the eigenvectors of the covariance matrix constructed from the data matrix. Using Singular Value Decomposition (SVD) (Gerbrands 1981) technically simplifies the calculations, which is an important issue in case of large data matrices. A score plot is the projection of the transformed data matrix on the plane of two arbitrarily chosen PCs. In this representation a spectrum is a point on the plane of the score plot. Clustering of such points shows similarity of the corresponding spectra. A LabView program was written for performing PCA on arbitrarily chosen set of spectra, and for displaying the desired score plots. Another LabView program projects arbitrary spectra on a previously constructed score plot. As it was found, the first 2–3 PCs account for more than 90% of the variance in the data set. The LIBS spectra were vector normalised before the PCA was performed on them. The changes in the overall emitted intensity that are due to the pulse-to-pulse fluctuations of the laser fluence, or to the slight changes of the

measurement geometry from sample to sample, are eliminated this way. It should be noted though, that this normalisation also reduces those differences between spectra, which occur in consequence of the differences in the laser-to-sample coupling efficiency. The advantage of using PCA together with a LIBS database is, that placing the spectrum of an unknown paint on the score plot of known set of samples would reveal the similarities or differences between the unknown material and the paints of our database. In the next section, the benefits and limitations of this method will be examined. 3

RESULTS

3.1 Differentiation between paints of similar colours, choice of database First of all, it was examined how sensitively we can differentiate different paints of similar colours by the combined LIBS-PCA technique. As an example, a score plot of six different greens is shown on Figure 1. The points of the same type belong to single-shot measurements on the same sample. The paints containing green earth, chromium oxide and viridian are easily distinguished from the group where the paints containing malachite, copper acetate and Rinmann’s green pigments are found. Rinmann’s green can also be well distinguished from malachite and copper acetate, if the data set of PCA is restricted to them. What is more, malachite and copper acetate, differing only in stoichiometry of Cu, C, O, H also tend to separate when the PCA data set consists only of these

Figure 1. PCA score plot of different green paints. Clusters of green earth (G1_6) viridian (G9_6) and chromium oxide (G8_6_a,b) containing paints can be well distinguished from those of Rinmann’s green (G7_6), malachite (G3_6), and copper acetate (G4_6) paints.

151

LACONA8.indb 151

11/10/2010 5:37:05 PM

Figure 2. Score plot of malachite (G3_6) and copperactetate (G4_6) containing paints.

in the spectrum of the aged paint, while it is more present in the spectrum of the touch dry paint (Figs. 3–4). The difference in the spectra of fresh and aged paints was followed up in time, by repeating the measurements on a few day timescale. As Figure 5 shows, the spectra of the 9 years old paint (G8_8_a,c) taken on 24.08.2009. and 28.08.2009. respectively, fall in the same cluster, although the deviation is relatively high, most probably due to the inhomogeneity of the layer. (Brush strokes are clearly visible in it). The freshly painted chromium oxide (G15_a,b,c) paint spectra separate from the cluster of the old paint, and there seems to be some clustering within them according to measurement dates (24.08.2009, 25.08.2009, 28.08.2009, respectively).

two paints (Fig. 2). By their LIBS spectra, malachite and copper acetate were not distinguishable to the eye. A new spectrum (measured with the standard parameters given in 2.2) can be placed on the database score plot by a simple projection onto the known PCs. As a trial, a new, independent set of measurements was performed on the chromium oxide containing aged paint, and the resulting spectra were placed on the score plot (Fig. 1, G8_6_b). The similarity is convincing. 3.2

The influence of the binding material

The possibility of differentiation between paints containing the same pigment and different binding materials was investigated by means of two series of samples made with Prussian blue and cobalt blue pigments, respectively, and with 10 different binding materials (Table 1). The samples were freshly prepared. Only paints with egg yolk binding could be separated from the others, which is due to the relative abundance of iron, phosphorus, calcium and zinc in the yolk. In general, it can be concluded that the presence of the examined fresh binding materials (except egg yolk) does not influence significantly the LIBS spectra of paints containing Prussian blue and cobalt blue pigments. For further testing, a new sample, consisting of the same chromium oxide pigment and the same binding material (linseed oil) as sample G8_8 was prepared, and spectra were taken at different intervals of time with respect to the date of preparation. Interestingly, these spectra were quite different from those of the several years old paint of the same composition. Comparing them to the spectrum of chromium reference (chromium pieces, 99.99% Alfa Aesar®), it was found that a high intensity, wide band emission component, which is characteristic of pure chromium, is decreased

Figure 3. LIBS spectra of the 9 years old paint sample (chromium oxide in linseed oil, G8_8), taken with 3 days time difference (averages of 10 single measurements).

Figure 4. LIBS spectra of freshly painted sample containing chromium oxide pigments in linseed oil binding, taken 1 and 4 days after touch drying (G15_a, G15_b, respectively), and the LIBS spectrum of the 99.99% clear chromium reference (averages of 10 single measurements).

152

LACONA8.indb 152

11/10/2010 5:37:06 PM

Figure 5. Score plot of the LIBS spectra of 9 years old (G8_8) and freshly painted (G15) samples containing the same chromium oxide pigments and linseed oil binding, taken with several day time differences (a: 24.08.2009, b: 25.08.2009, c: 28.08.2009).

Figure 7. PCA score plot of paints containing linseed oil as binding material and cobalt blue: Prussian blue pigments in 100:0, 85:15, 50:50, 27:63, 0:100, 96:4 w/w% composition for samples B100, B85, B50, B27, B0, B96 respectively.

3.3

Figure 6. LIBS spectra of the 9 years old G8_6 paint sample (chromium oxide pigments in rabbit skin glue), taken with 3 days time difference (averages of 10 single measurements).

The possible explanation of the above phenomenon is that polymerisation and cross-linking take place during this early stage of oil aging, leading to significant changes in the UV absorption of the oil film (Boyatzis 2002, Groza 2004). Based on the above observations, one can conclude that chromium, having intensive emission in the UV, can serve as a good indicator of the aging of the binding material. This suspect is confirmed by the LIBS spectra of the 9 years old sample containing chromium oxide pigments in rabbit skin glue (Fig. 6). The absorption of the aged glue in the UV is even more apparent than that of aged linseed oil, by comparison of Figures 3, 4 and 6. Further, systematic experiments are planned to verify the applicability of chromium containing pigments as indicators of curing of different binding media.

Mixtures of diverse concentrations

Samples B0-B100 contain cobalt and Prussian blue pigments in linseed oil. For each sample 100 units of pigments in different w/w% composition (Table 1) was mixed in the same amount of linseed oil. The mixture was homogenized with a magnetic stirrer. The paint was applied on the same ground as the previous samples. As it is shown on Figure 7, PCA can differentiate between the paints according to the cobalt blue: Prussian blue pigment ratio. The spectra of the B96 sample were projected on the score plot of the other samples. They fit convincingly in. Although this method presumes standards, its simplicity and applicability in case of highly heterogeneous samples, like the paint mixtures under consideration make it competitive with calibration-free LIBS (CF-LIBS) method (Fotakis 2007, p. 65). The available accuracy of the LIBS-PCA method in the determination of percentage composition of pigment mixtures shall be examined in further experiments. 4

SUMMARY

Potential of laser induced spectroscopy in identification of complex paint layers was investigated in the present study. A wide-range, high-resolution spectrograph collected the spectra, and principal component analysis was used for processing the large amount of collected data. It was shown on an example that PCA can differentiate between spectra of similar pigments— that are difficult to distinguish otherwise by their LIBS spectra—even if they are examined not in

153

LACONA8.indb 153

11/10/2010 5:37:06 PM

their clear form, but as components of a paint layer. The efficiency of this discrimination depends on the choice of database. Spectra of a range of freshly prepared paints containing Prussian blue and cobalt blue pigments and different (proteinaceous, polysaccharide and fatty acid) binding materials were found to be indistinguishable with combined LIBS-PCA, except when the binding material was egg yolk. However, evidences were found, that—due to their intensive emission in the UV—chromium-containing pigments can be used for indication of the aging of the binding materials in paints. Different w/w percentage composition mixtures of cobalt and Prussian blue pigments, all applied with linseed oil binding, could be separated by the PCA score plot very well. In summary, it was found that PCA does contribute to the interpretation of LIBS measurements. Furthermore, careful considerations regarding the matrix effect in the course of testing complex materials with LIBS, can lead to conclusions beyond identification of elemental lines. The simple and sensitive combined LIBS-PCA technique can be useful in paint identity authentication problems as a first step of the analytic procedure. It can exclude identity of two samples, or give indications as to which further analytic methods to use. ACKNOWLEDGEMENTS Zs. Márton is grateful to P. Heszler for fruitful conversations. This paper is devoted to his memory. Furthermore, the financial support of National Office for Research and Technology under grant No. 2006ALAP3-00445/2006 is acknowledged. Thanks to F. Kaposvári for XRF analysis. REFERENCES Boyatzis, S. et al. 2002. UV exposure and temperature effects on curing mechanisms in thin linseed oil films: Spectroscopic and chromatographic studies. Journal of applied polymer science 84(5): 936–949. Brundle, C.R. et al. 1992. Encyclopedia of materials characterization. Oxford: Butterworth-Heinemann.

Chiavari, G. et al. 1998. Characterisation of standard tempera painting layers containing proteinaceous binders by pyrolysis (/methylation)-gas chromatography-mass spectrometry, Chromatographia, 47(7/8): 420–427. Ciliberto, E. & Spoto, G. 2000. Modern analytical methods in art and archaeology, Chemical analysis, monographs on analytical chemistry and its applications. Vol. 155. New York: John Wiley. Cremers, D.A. & Radziemski, L.J. 2006. Handbook of Laser-Induced Breakdown Spectroscopy. New York: John Wiley. Diwakar, P.K. et al. 2007. The effect of multi-component aerosol particles on quantitative laser-induced breakdown spectroscopy: Consideration of localized matrix effects. Spectrochimica Acta B 62: 1466–1474. Eastaugh, N. et al. 2005. Pigment compendium. CDROM. Oxford: Butterworth-Heineman. Fotakis, C. et al. 2006. Lasers in the preservation of cultural heritage; Principles and applications. New York: Taylor & Francis. Gerbrands, J.J. 1981. On the relationships between SVD, KLT and PCA, Pattern Recognition 14(1–6): 375–381. Groza, A. et al. 2004. Infrared spectral investigation of the linseed oil polymerization in a corona discharge in air at atmospheric pressure. Europhys. Lett. 68: 652–657. Huang, J.S, et al. 2004. Matrix effect on emission current correlated analysis in laser-induced breakdown spectroscopy of liquid droplets. Spectrochimica Acta B 59: 321–326. Jolliffe, I.T. 2002. Principal component analysis. New York: Springer-Verlag. Nevin, A. et al. 2008. The analysis of naturally and artifically aged protein-based paint media using Raman spectroscopy combined with Principal Component Analysis. Journal of Raman Spectroscopy, 39: 993–1000. Osticioli, I. et al. 2008. An Optimization of Parameters for Application of a Laser-Induced Breakdown Spectroscopy Microprobe for the Analysis of Works of Art. Applied Spectroscopy 62(11): 1241–1249. Osticioli, I. et al. 2009. Analysis of natural and artifical ultramarine blue pigments using laser indiced breakdown and pulsed Raman specroscopy, statistical analysis and light microscopy. Spectrochimica Acta A 73: 525–531. Schroeder, D.J. & Hillard, R.L. 1980. Echelle efficiencies: theory and experiment. Applied Optics, 19(16): 2833–2841.

154

LACONA8.indb 154

11/10/2010 5:37:07 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Study of matrix effect in the analysis of pigments mixtures using laser induced plasma spectroscopy M.P. Mateo, T. Ctvrtnickova, A. Yañez & G. Nicolas Laboratorio de Aplicaciones Industriales del Láser, Universidad de A Coruña, A Coruña, Ferrol, Spain

ABSTRACT: In authentication and restoration tasks of painted artworks, it is crucial to know if the layer under study corresponds to a pure pigment or to a mixture, and, in the second case, in which proportion. Laser induced plasma spectroscopy has been extensively used for the analysis of pigments in artworks and archaeological pieces. In this work, the effect of matrix composition in the analysis of pigments mixtures using laser induced plasma spectroscopy is studied. For this purpose, pure Zinc White and Ultramarine Blue pigments and mixtures of these two pigments prepared in different proportions were analyzed. Spectral results of mixtures have been correlated to pigments concentrations to check if there exist interferences due to matrix composition or if there is a direct correlation between intensities and concentrations. Although matrix effect was recognized in zinc emission lines, it was corrected by normalizing the signals of the two pigments. In this way, a good correlation between spectral signal and pigments concentration was obtained. 1

INTRODUCTION

The identification of pigments used in an artwork is crucial for the artistic characterization, including dating, and also for the restoration tasks of the painted work. Among other techniques more established, Laser induced plasma spectroscopy (Cremers & Radziemski 2006), Laser-Induced Plasma Spectroscopy (LIPS) also known as LIBS, has been extensively used during the last decade for the analysis of pigments in artworks and archaeological pieces (Anglos 2001; Anglos & Miller 2006; Giakoumaki et al., 2007), including icons (Burgio et al., 2001; Osticioli et al., 2008), wall and easel paintings (Brysbaert et al., 2006; Bruder et al. 2007), polychromes (Castillejo et al. 2000; Lopez et al., 2008), and illuminated manuscripts (Bicchieri et al., 2001; Melessanaki et al., 2001). LIPS data about pigments composition has been used for different purposes: dating the artwork on the basis of the materials and the pictorial technique used, providing information about their source, establishing the state of preservation, distinguishing original artworks from falsifications, recognition of pigments of the same color with different composition in order to select the adequate method for restoration, etc. LIPS analysis consists of the acquisition and processing of an emission spectrum from a laser ablated material and resulting plasma. This spectrum can be used like a fingerprint of the pigment. In this sense, the identification of

the characteristic emission peaks in the spectrum allows the discrimination between different pigments (Anglos 2001). Although the information obtained from the spectral analysis is elemental, in many cases the elements identified in the spectrum enable the recognition of the pigment present in the work of art. In authentication and restoration tasks of painted artworks, it is very important to know if the layer under study corresponds to a pure pigment or to a mixture, and, in the second case, in which proportion. This information can be crucial to ensure that the same pigments that were used in the original work are employed for the restoration and therefore reproduce the same color of the original piece. In this work, the effect of matrix composition in the analysis of pigments mixtures using LIPS is studied. For this purpose, mixtures of two pigments, Zinc White and Ultramarine Blue were prepared in different proportions and the corresponding laser induced plasma signal was acquired. Pure pigments were also analyzed for comparison. Spectral results of the prepared mixtures have been correlated to pigments concentrations to check if there exist interferences due to matrix composition or if there is a direct correlation between intensities and concentrations. A good correlation has been obtained after data processing to correct matrix effects. In addition, the influence of the binding agent concentration on the LIPS analysis of a pure pigment has been also studied.

155

LACONA8.indb 155

11/10/2010 5:37:07 PM

2

EXPERIMENTAL

2.1

3

Set up

3.1

A pulsed Nd:YAG laser operating at its second harmonic 532 nm, with a pulse energy of 30 mJ and a pulse duration of 5 ns was used in combination with an Echelle spectrograph (spectral region of 200–850 nm) and an ICCD camera detector. Additional information about the experimental set up used can be found elsewhere (Mateo et al., 2009). Temporal parameters were: delay time 300 ns and integration time 10 μs. LIPS signal from 9 pulses (each on a fresh sample surface) was averaged. 2.2

Sample preparation

Two set of samples were prepared. To study the influence of the binding agent, Zinc White (ZnO) pigment powder was mixed with a binding agent commonly used in painting, linseed oil, in different ratios: 34/66, 38/62, 42/58, 46/54, 50/50, 54/46, 58/42, 62/38 and 66/34. For the study of the matrix effect of pigment mixtures, Zinc White and Ultramarine Blue (Lapis Lazuli—Na8–10 Al6Si6O24S2–4,) were used. These pigments were mixed in their powder form in specified ratios: 20/80, 40/60, 60/40 and 80/20. These mixtures and the two pure pigments were bound separately with linseed oil in a 1:1 ratio. All the samples were spread in a thick layer on wooden plates and were dried. Figure 1 shows a typical LIPS spectrum of Zinc White and Ultramarine Blue. These two pigments were selected because their emission lines are easy to recognize and do not interfere, as can be noted in the spectra. As shown, Zinc White spectrum is very simple with Zn emission lines, while Ultramarine Blue has more complex spectrum containing Ca, Al, Na, Li, K emission lines. These were identified in the spectra using a software developed in our lab (Mateo et al., 2005). Al

Intensity (Couns)

20000 0 40000

Paints are constituted of pigments dispersed in a binding medium and this mixture is applied on the surface to be painted. In a preliminary study, the influence of the ratio between pigment powder and binding agent in the LIPS signal of the pigment was studied. For this purpose, Zinc White and linseed oil were selected as typical pigment and binder used in art, respectively. Both were mixed in different ratios and the corresponding samples were analyzed by LIPS. The results obtained are shown in Figure 2, where the peak area corresponding to Zn (I) 334.502 nm emission line has been plotted against the percentage of Zinc White in the mixture pigment-binding agent. The rectangle of Figure 2 includes the points corresponding to the concentration range where the signal is stable and the reproducibility is acceptable. In this sense, Figure 2 reveals that the suitable amount of added oil binder to Zinc White pigment falls within the range of 45–60%. It is evident, that less than 40% of binder addition causes high fluctuation of signal, while small variation around 50% does not affect significantly the LIPS signal. This result agrees with the common mixing ratio of pigment with binding agent used in artworks, that is 1:1. It should be noted that percentages higher than 66% of linseed oil were not included in the study due to the difficulty for the drying process observed, which make them inappropriate for artistic purposes. 3.2

In this study, the effect of matrix composition in the analysis of pigments mixtures using LIPS was % of linseed oil 65

60

55

50

45

40

35

40

45

50

55

60

65

5

8.0x10

Ca Ca Na

Al Na 300

Correlation between LIPS signal and pigment ratio

a)

Na

80000 60000

Influence of binding agent

Peak area (a. u.)

100000

40000

RESULTS AND DISCUSSION

Li 400

Zn

Ca 500

600

Li

K 700

800

b)

30000

5

6.0x10

5

4.0x10

5

2.0x10

0.0

20000

35

Zn

10000

Zn

% of Zinc White pigment

0 300

400

500

600

700

800

Wavelength (nm)

Figure 1. LIPS spectra obtained from the two pure pigments under study: a) Ultramarine Blue, b) Zinc White. Signal corresponding to nine pulses was averaged.

Figure 2. Plot of the LIPS peak area corresponding to the Zn (I) 334.502 nm emission line versus concentrations of linseed oil and the concentration of Zinc White pigment in the mixture of both. Error bars correspond to nine measurements.

156

LACONA8.indb 156

11/10/2010 5:37:07 PM

investigated. For this purpose, mixtures of two pigments, Zinc White and Ultramarine Blue, were prepared in different ratios and the corresponding laser induced plasma signal was acquired. Pure pigments were also analyzed for comparison. Figure 3 shows the spectral and visual aspect of the paint mixtures at different ratios of white and blue pigments. Only a portion of the total spectra has been shown for clarification. As expected, pigment mixtures with a high content of Ultramarine Blue present an intense blue color and produce spectra where Al emission lines predominate. In contrast, those mixtures with high content in white pigment exhibit a faded color and are characterized by intense Zn emission lines. Spectral results

of these mixtures were correlated to pigments concentrations to check if there exist interferences due to matrix composition or if there is a direct correlation between intensities and concentrations. Hence, peak area of main emission lines of characteristic elements was plotted against pigment concentration in the paint mixture. The results corresponding to characteristic elements of Ultramarine Blue pigment are shown in Figure 4. As expected, the LIPS signal of Si, Al, Ca, K, Mg and Na from blue pigment increases with pigment concentration, demonstrating that a good correlation between LIPS signal and blue pigment concentration exists. The general trend of the dependence is exponential.

Al

a)

Zn

b)

c)

d)

e) f) 300

305

310

315

320

325

330

335

340

Wavelength (nm) Figure 3. LIPS spectra of the paint mixture at different ratios of Zinc White and Ultramarine Blue pigments: a) 100% white, b) 20% blue + 80% white, c) 40% blue + 60% white, d) 60% blue + 40% white, e) 80% blue + 20% white, f) 100% blue. Only a portion of the spectra is shown for clarification. Each spectrum corresponds to 1 laser pulse. The photographs of the corresponding sample surfaces are also shown.

157

LACONA8.indb 157

11/10/2010 5:37:08 PM

5

Peak area (a. u.)

6.0x10

d)

a)

5

6.0x10

5

5

4.0x10

5

2.0x10

4.0x10

5

2.0x10

0.0

0.0 5

0

20

40

60

80

100

0

4x10

40

60

80

100

5

2.0x10

e)

b) Peak area (a. u.)

20

5

5

1.5x10

5

1.0x10

5

5.0x10

3x10

5

2x10

4

1x10

0.0

0 5

1.0x10

0

20

40

60

80

100

0

20

40

60

80

100

f)

c)

6

6.0x10

Peak area (a. u.)

4

8.0x10

6

4.0x10

4

6.0x10

4

4.0x10

6

2.0x10

4

2.0x10

0.0

0.0 0

20

40

60

80

100

0

% of Blue pigment

20

40

60

80

100

% of Blue pigment

Figure 4. Plot of the peak area of a) Si (I) 288.158 nm, b) Ca (I) 422.673 nm, c) Mg (I) 285.213 nm, d) Al (I) 309.271 nm, e) K (I) 766.49 nm and f) Na (I) 589.592 nm emission lines in LIPS spectra, versus concentration of blue pigment. Error bars correspond to nine measurements.

However, an unexpected behavior of Zn spectral line was obtained and can be observed in Figure 5. The intensity (peak area) of Zn (I) 334.502 nm was growing when decreasing the concentration of the white pigment in the paint mixture. This feature was assigned to a matrix effect, that is, Zinc White pigment present a different ablation behavior when it is alone than when it is mixed with Ultramarine Blue. This effect can be observed in Figure 5, where the point corresponding to 100% of white pigment does not follow the trend of the rest of points. On the other hand, one can expect that the LIPS signal peak area corresponding to the zinc emission line will increase with the concentration of Zinc White

in the paint mixture. However, on the contrary, Figure 5 shows that when the concentration of blue pigment increases in the mixture, the ablation is more effective and therefore the overall spectrum of the mixture increases in intensity, including the zinc emission lines, although the concentration of the white pigment decreases. In this case, zinc peak area can not be properly correlated with white pigment concentration due to a matrix effect. In order to surpass the matrix effect and to obtain linear calibration dependence, signal normalization was applied. In this sense, for each sample, zinc peak areas were normalized to silicon peak areas and plotted

158

LACONA8.indb 158

11/10/2010 5:37:09 PM

4

% of White pigment 5

100

80

0

20

60

40

20

0

40

60

80

100

1.4x10

5

Peak area (a. u.)

1.2x10

5

1.0x10

4

8.0x10

4

6.0x10

4

4.0x10

4

2.0x10

0.0

% of Blue pigment Figure 5. Plot of the peak area of Zn (I) 334.502 nm emission line in LIPS spectra versus concentrations of blue and white pigments. Error bars correspond to nine measurements.

% of White pigment 80

60

40

20

0

Zn/Si peak area ratio

1.2

0.8

0.4

R = 0.9898

0.0 20

40

60

80

100

% of Blue pigment Figure 6. Plot of the peak area ratios corresponding to Zn (I) 334.502 nm and Si (I) 288.158 nm emission lines versus concentrations of blue and white pigments. Error bars correspond to nine measurements. Solid line corresponds to a linear fit while dotted lines show the confidence bands.

CONCLUSIONS

This is a preliminary study of the effect of matrix composition in the analysis of pigments mixtures using laser induced plasma spectroscopy. First, the influence of the concentration of the binding agent, linseed oil, on the LIPS signal characteristic of Zinc Whit pigment was studied. The results revealed that, when the linseed oil is in a concentration range between 40 and 60%, the LIPS pigment signal is stable and the reproducibility is acceptable, which is in agreement with the ratio 1:1 usually used by artists in their mixtures of pigments with binding agents. In order to study the matrix effect, LIPS analyses were performed in mixtures of Zinc White and Ultramar Blue in different ratios and also in the pure pigments. The correlation curves between LIPS signal and concentration of pigment corresponding to the elements characteristic of Ultramarine Blue, Si, Ca, Al, Mg, K and Na, exhibited an exponential trend, increasing the peak area when the pigment. concentration was higher. However, matrix effect was found when plotting the LIPS signal corresponding to Zn (I) 334.502 nm emission line, characteristic of the white pigment. In this case, although the Zinc White concentration increases, the peak area of the zinc lines decreases. In order to correct this effect, zinc signal was normalized to the signal of the elements characteristic of Ultramarine Blue. After this data processing, linear correlation was found between normalized LIPS peak area and concentration of pigment. This good correlation could be used by artists and restorers to estimate from the LIPS analysis the proportion in which two pigments were mixed by the author in an artwork. More work needs to be done about this regard, studying other pigments and mixtures of more than two components.

REFERENCES

against concentration of the pigment in the paint mixture. The results obtained are shown in Figure 6. As opposed to the behaviour exhibited in Figure 5, Zn normalized signal increases when the concentration of Zinc White increases in the mixture, and even presents a linear correlation with a R = 0.9898. It should be noted that this was not an isolated case, that is, a similar behaviour was observed when normalizing zinc peak area to the corresponding of other characteristic elements of Ultramarine Blue pigment: Ca, Al, Mg, K and Na.

Anglos, D. 2001. Laser-induced breakdown spectroscopy in art and archaeology. Applied Spectroscopy 55(6): 186 A–205 A. Anglos, D. & Miller, J.C. 2006. Cultural heritage applications of LIBS. In Miziolek, A.W., Palleschi, V. & Schechter, I. (eds), Laser Induced Breakdown Spectroscopy (LIBS): fundamentals and applications: 332–367. Cambridge: Cambridge University Press. Bicchieri, M., Nardone, M., Russo, P.A., Sodo, A., Corsi, M., Cristoforetti, G., Palleschi, V., Salvetti, A. & Tognoni, E. 2001. Characterization of azurite and lazurite based pigments by laser induced breakdown spectroscopy and micro-Raman spectroscopy. Spectrochimica Acta Part B-Atomic Spectroscopy 56(6): 915–922.

159

LACONA8.indb 159

11/10/2010 5:37:10 PM

Bruder, R., Detalle, V. & Coupry, C. 2007. An example of the complementarity of laser-induced breakdown spectroscopy and Raman microscopy for wall painting pigments analysis. Journal of Raman Spectroscopy 38(7): 909–915. Brysbaert, A., Melessanaki, K. & Anglos, D. 2006. Pigment analysis in Bronze Age Aegean and Eastern Mediterranean painted plaster by laser-induced breakdown spectroscopy (LIBS). Journal of Archaeological Science 33(8): 1095–1104. Burgio, L., Melessanaki, K., Doulgeridis, M., Clark, R.J.H. & Anglos, D. 2001. Pigment identification in paintings employing laser induced breakdown spectroscopy and Raman microscopy. Spectrochimica Acta Part B-Atomic Spectroscopy 56(6): 905–913. Castillejo, M., Martin, M., Silva, D., Stratoudaki, T., Anglos, D., Burgio, L. & Clark, R.J.H. 2000. Analysis of pigments in polychromes by use of laser induced breakdown spectroscopy and Raman microscopy. Journal of Molecular Structure 550: 191–198. Cremers, D.A. & Radziemski, L.J. 2006. Handbook of laser-induced breakdown spectroscopy. New York: Wiley. Giakoumaki, A., Melessanaki, K. & Anglos, D. 2007. Laser-induced breakdown spectroscopy (LIBS) in archaeological science-applications and prospects. Analytical and Bioanalytical Chemistry 387(3): 749–760.

Lopez, A.J., Mateo, M.P., Santaclara, A. & Yanez, A. 2008. Compositional Analysis of Polychromes by Laser-Induced Breakdown Spectroscopy. Materials Science Forum 587–588: 657–661. Mateo, M.P., Ctvrtnickova, T. & Nicolas, G. 2009. Characterization of pigments used in painting by means of laser-induced plasma and attenuated total reflectance FTIR spectroscopy. Applied Surface Science 255(10): 5172–5176. Mateo, M.P., Nicolas, G., Pinon, V., Alvarez, J.C., Ramil, A. & Yanez, A. 2005. Versatile software for semiautomatic analysis and processing of laser-induced plasma spectra. Spectrochimica Acta Part B-Atomic Spectroscopy 60(7–8): 1202–1210. Melessanaki, K., Papadakis, V., Balas, C. & Anglos, D. 2001. Laser induced breakdown spectroscopy and hyper-spectral imaging analysis of pigments on an illuminated manuscript. Spectrochimica Acta Part B-Atomic Spectroscopy 56(12): 2337–2346. Osticioli, I., Wolf, M. & Anglos, D. 2008. An Optimization of Parameters for Application of a LaserInduced Breakdown Spectroscopy Microprobe for the Analysis of Works of Art. Applied Spectroscopy 62(11): 1242–1249.

160

LACONA8.indb 160

11/10/2010 5:37:10 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Pomerania Laboratory—A solution for the cultural heritage research and conservation A. Iwulska, I. Traczyńska, R. Jendrzejewski, M. Sawczak & G. Śliwiński Photophysics and Laser Technique Department, Polish Academy of Sciences, IF-FM, Gdańsk, Poland

A. Kriegseisen National Museum, Gdańsk, Poland

ABSTRACT: Results of the investigation collected so far for the XVI c., gilded figure presenting the St George killing the dragon, and obtained by means of the experimental capacity available at the Pomerania Laser Laboratory in Gdańsk are presented in this paper. The complementary use of the elemental (XRF, LIPS and EDS) and structural analysis revealed, that two gilding techniques: with the usage of gold and of the Zn-Cu alloy, were applied on different parts of the figure in the past. From the μ-Raman spectra the beeswax coverage of the entire surface was concluded, as well as the main components of the surface soiling: soot (C), microcrystalline CaCO3, and the corrosion product Cu3SO4(OH)4 (antlerite) of the copper figure were identified. The stratigraphic data of the protective layers obtained by means of laser ablation confirmed the presence of the ∼50 μm thick gilding with Au layer on some figure parts, in accordance with the XRF and LIP spectra. Irradiation by single laser pulses (0,3 J/cm2) applied for detachment of the contaminated beeswax layer and followed by its mechanical removal resulted in the successful surface cleaning. 1

INTRODUCTION

During the last two years the project “Diagnostic Techniques and Optoelectronics for Conservation” supported by the EC Program “Competitiveness”, via the European Fund for Regional Development, was realized at the Pomerania Laser Laboratory, Polish Academy of Sciences, IF-FM in Gdańsk. The project was aimed on completion of, and access to the up-to-date equipment for the research and conservation projects on cultural heritage in the region. In frames of the project the existing experimental capacity was completed. It provides, together with the extensive multidisciplinary collaboration, networking and qualified personal, the complete, high level service offered to researchers and conservators. First evaluation of the project outcomes indicate that the field applications of the transportable equipment such as the laser cleaning stations, the X-Ray Fluorescence (XRF) and Laser-Induced Plasma (LIPS) spectrometers and also the 3D laser scanner represent the most frequently exploited equipment. The advantages of complementary use of the Raman and XRF techniques for analysis of the historical paint layers were discussed in our previous paper (Sawczak et al. 2009). Also the potential and sensitivity of the LIPS technique applied for

the structural and chemical characterization of historical objects is well described in the literature (Oujja et al. 2005). In this work, on the example of the recently performed conservation project of the XVI c., allmetal, gilded figure of natural height presenting the St George killing the dragon (Fig. 1), results regarding the laser surface cleaning, the object analysis and documentation aimed on the choice of the conservation strategy are presented and discussed.

Figure 1. The figure of St. George killing the dragon, XVI c.; a photograph of the complete figure (a), and main elements separated for conservation (b, c).

161

LACONA8.indb 161

11/10/2010 5:37:10 PM

2

THE OBJECT—ST. JOHN KILLING THE DRAGON

The object under conservation has been located on the St. George’s Confraternity Court tower in Gdansk in 1566 and remained there till 1945 when the building was destroyed (Friedrich et al. 1997). In the period of XV-XVII c., the Court belonged to the most prestigious part of the Gdansk’s society. The original figure of the total surface of ca. 5 m2 was made of riveted, gilded copper elements. During previous conservation interventions some damaged parts were substituted by their counterparts produced of copper, too. Moreover, the exposed figure surface was covered with the wax layer in post-war years. Recently, the object is investigated at the National Museum in Gdansk. Analyses indicate serious contamination of the entire figure surface and also damaged areas of the protective layers and local corrosion spots due to the cumulative environmental effect. Some minor parts of the figure are missing and require reconstruction. The research and conservation should result in placement of the figure on the Court tower again. 3

EXPERIMENTAL

For the object documentation and restoration purposes, the high-resolution scanning of the entire object by means of a portable 3D laser scanner (VI-9i, Konica-Minolta) was performed. The measurement inaccuracy was equal to +/− 0.3 mm and the single scan duration for collection of the data matrix of surface shape and color was 2, 5 s. The laser cleaning was performed by means of the pulsed Q-switched Nd:YAG laser (Laserblast 1000, Quantel) operated at 1064 nm, and of pulse length and repetition rate of 10 ns and 120 Hz, respectively. The beam intensity distribution was homogeneous (“top-hat” profile) over the controllable beam cross-section of a square shape. The non-destructive sampling of the object elemental composition was performed in-situ by means of the portable, high-sensitivity XRF spectrometer completed at IF-FM PASci. The spectrometer X-ray tube (IS601.5, Italstructures) used for excitation produced the beam collimated to a spot of 4 mm in diameter. The energy resolution of the detection system (AXAS, Ketek) of 155 eV (Mn Kα line, 5.9 keV) and the detection limits of 50–340 ppm (Cu-Pb) were confirmed during the instrument calibration. The tube was typically operated at 55 kV and 1 mA, and without He shielding it assured an efficient excitation of the Kα lines of elements of the atomic number Z > 19. The accumulation time of 120 s for each spectrum was applied.

On several areas of the protective layers, the structure and elemental composition were investigated by means of the micro-ablative material penetration accompanied by recording of the Laser Induced Plasma spectra (LIPS), too. The sample irradiation and excitation was provided by the Nd:YAG laser (Quantel B) pulses at 1064 nm (8 ns). The pulse energy density on the sample surface was equal to 3,3 J/cm2 and the laser beam was carefully focused in order to assure minimal destruction. For recording of the LIP spectra the plume emission was analyzed by a 0.3 m spectrograph (SR-303i, Andor Technologies) equipped with a 600 grooves/mm grating and coupled to the timegated ICCD camera (DH 740, Andor Technologies). Spectra were recorded using the time integration mode, in the wavelength range from 275 to 650 nm in steps of 40 nm at a resolution of 0,3 nm, and with a gate width and delay of 10 μs and 1 μs, respectively. SEM analyses were carried out by means of the Carl Zeiss instrument (LEO 1430 VP) coupled with the EDS apparatus (Quantax 200 & detector XFlash 4010). The complementary measurements of the chemical composition of the object were performed by means of the μ-Raman spectrometer. For this purpose small samples (∼few mg) from different parts of the St. George and dragon figures, including these covered by wax and the corrosion and contamination layers were extracted. Sampling was performed only at locations previously investigated with the XRF instrument. The confocal spectrometer (InVia, Renishaw) providing laser excitation of the samples at 785 nm was applied. The objective of the magnification of 50 × assured the spectral footprint of the sample surface area of about 5 × 5 μm2. Spectra were acquired over the range of 100–3200 cm−1 at resolution of 2 cm−1. Depending on the signal intensity the spectra of individual samples were accumulated and averaged to improve the signal/noise ratio. 4 4.1

RESULTS AND DISCUSSION Digital documentation

The 3D laser scanning was performed for the entire surface of the disassembled object comprising 5 groups of the figure parts, and the total number of scans made was about 80. Next, the single scans were processed and assembled by means of the specialized RapidFormXO™ software. The mechanically damaged and also missing parts of the figure were reproduced digitally basing on photographs and dimensions of similar parts or these showing symmetry. An example of the reconstruction result of the missing finger of the St. George’s left hand is shown in Figure 2.

162

LACONA8.indb 162

11/10/2010 5:37:11 PM

Cu

exposed side reverse side reverse side without protective layers

5

Cu

Pb

Sb

Sb

Ba

Pb Au

3

10

Pb

Fe

4

10

Fe

Intensity [a.u.]

10

Au

In this case the corresponding digital data of the right hand were used. The 3D polygonal mesh of the appropriate finger was separated, transformed by the mirror-like reflection and attached to the location of the missing part. The ready to use, digital representation of the complete figure occupies the total volume of about 47 GB and the demonstration in frames of a virtual museum is in progress.

Ca Ba Ba

Figure 2. Result of the 3D laser scanning applied for reconstruction of the missing part of the figure; here— the left hand finger.

Figure 3. The laser cleaning of elements of the St. George figure: helmet (a), and the head (b)—comparison of the cleaned and non-cleaned surface parts.

2

10

1

10

4.2

Laser cleaning

10

The aim of the laser cleaning was to remove the soiling from the figure surface, without destruction of the gilding layers. The surface observation indicated that the upper protective layer of organic origin— most probably wax, is partially missing or damaged in some areas and the remaining part is penetrated by contaminants due to prolonged influence of the environmental pollution. For this reason the complete removal of the organic layer and soiling was decided by means of the laser ablation. The presence of the reflecting surface of the gilding under the contaminated layer which was partially transparent to the laser radiation assured selective absorption of the irradiation energy. During experiment the constant energy density of 300 mJ/cm2 was applied. The contaminated layer was detached from the gilding by irradiation with single shots on the neighboring spots of the area of 7 × 7 mm2 at low pulse frequency of 2 Hz. The cleaning process was completed by the mechanical removal of the detached layer. It was found, that such combination of the laser cleaning followed by the mechanical removal of the remnants lead to the best final result. For comparison, the cleaned and non-cleaned parts of the surface are shown on the right hand side photograph in Figure 3. 4.3

Elemental composition of the figure surface

The XRF spectra were taken from several parts of the figure and these representative for the dragon’s

20 Energy [keV]

30

Figure 4. The XRF spectra obtained for the figure body—dragon paw, on the exposed (solid line) and reverse side (thin solid line), and for surface area not covered by the protective layers (dashed line).

paw on the exposed (solid line) and reverse side (thin solid line), and also for the area not covered by the protective layers of gilding and wax (dashed line) are shown in Figure 4. The data are not corrected for the background containing the Zr, Ag and Sn lines and originating from the internal collimator of the detector. Elements such as Ca, Ba, Fe, Cu, Au, Pb, Sb are detected in the spectra and the observed line intensities confirm that main component of the figure material is copper while other elements represent about 3% wt of the material. Peaks corresponding to Au are observed both for the exposed and reverse side only in case of the surface covered by protective layers. Similar observation refers to elements Ba and Ca, which are not detected in area without wax and gilded layers. The Ca peak may derive from calcite, whose presence requires complementary analysis. The presence of Pb accompanied by traces of Ba and that of Fe may indicate on different gilding techniques applied in the past. The summary of the XRF data is given in Figure 5, where peak intensities related to elements present in the spectra of the exposed (surface 1) and reverse

163

LACONA8.indb 163

11/10/2010 5:37:11 PM

Ca Fe Zr Ag Sn Sb Ba Au Pb

Intensity [a.u.]

160

120

150

Intensity [ a.u. ]

200

80

100

Au I (392,77 nm) Au I (422,78 nm)

50 40

0

surface 1

surface 2

0

surface 3

0

Figure 5. Summary of the peak intensities of elements observed in the XRF spectra from Fig. 4; surfaces 1–3 correspond to the exposed, reverse, and non-protected parts of the figure, respectively.

Structure of the surface layers

In order to investigate the protective layers on the copper sheet the LIP spectra were recorded for consecutive laser ablation pulses applied to the same location on the surface. For detection of the Au element in the gilding the emission lines at 392.77 and 422.78 nm showing no coincidence with peaks of other elements were selected. A sharp decrease of the Au peak intensities is observed for both lines and 3 laser pulses are sufficient to remove the thin gilding completely—see Figure 6. This is evident from the absence of the Au signal for the following irradiation pulses and is confirmed by the SEM inspection of the irradiated area—insets in Figure 6. This together with the observed total thickness of ∼50 μm of the gilding allow to estimate the ablation rate of about 1 μm. The prolonged laser irradiation resulted in penetration of the primer layer and next of the copper sheet. This can be observed via the peak intensity dependence on the laser pulse number for

20

30

40

50

pulse number

Figure 6. Decay of the Au peak intensities from LIP spectra of the gilding vs consecutive laser pulse number; insets show the material penetration during analysis—on the left, and cross-section of the gilding—right hand side.

(surface 2) side, as well as that without wax and gilding (surface 3) are shown. Here, the element Cu dominating the composition was not taken into account. Comparison of the line intensities of Zr, Ag and Sn measured for surfaces 1, 2 and 3 provide estimate of the experimental inaccuracy of ∼25%. Admixtures of Ba are usually found in Fe2O3 and the nearly constant ratio of the Fe/Ba intensities together with the presence of Pb support the conclusion on different primer materials containing oxides Pb3O4 or Fe2O3, and eventually also two different gildings applied in the past. Traces of Au on surface 3 indicate local damage or almost complete removal of the protective coating. This is confirmed by the low intensity of the Pb, Fe and Ca lines measured for this surface. 4.4

10

30

Cu I (521,82 nm) Cu I (578,22 nm)

25

Intensity [ a.u. ]

20

15

10

5

0 0

20

40

60

80

100

pulse number

Figure 7. Changes of the Cu peak intensities vs laser pulse number observed extracted from LIP spectra during prolonged irradiation of the same spot on the figure surface.

the substrate spectral lines of Cu at 521,8 nm and 578,2 nm obtained from LIP spectra and shown in Figure 7. At the irradiation beginning (pulse number <10) the surface layers: soil, wax, gilding, primer and corrosion products are penetrated, stepwise removed, and the Cu substrate becomes uncovered which results in the regular intensity increase of both lines. Data obtained for the next pulses (<38) indicate signal instability which is due to removal of the primer layers and superposition of the effects of the sharp intensity distribution (∼Gaussian) of the irradiation spot, multiple reflections in the laser-drilled hole and explosive removal of the material remnants. These are completely removed due to the consecutive irradiation pulses (>40) and almost stable intensity is observed for the Cu bands.

164

LACONA8.indb 164

11/10/2010 5:37:11 PM

20 15

Cu I

Pb

Fe I

10

Cu I

Fe

5

Fe I

Intensity [ a.u.]

25

Fe II 0 50

pu lse nu mb er

35

15

300

3 1

350

400

450

500

λ [ nm ]

550

600

Figure 8. The LIP spectra obtained during ablative penetration of the surface layers and corresponding to the 1st, 15th, 35th and 50th laser pulse; the Pb, Fe and Cu bands correspond to the primer layer and substrate material with corrosion products, respectively.

Further data on the structure and composition of the surface layers follow from observation of other elements present in the LIP spectra. The overview of the line intensity changes in the spectral region of 300–625 nm in dependence on the in depth position probed by the laser beam reveals among Cu the peaks ascribed to Pb, and Fe—Figure 8. Changes in intensity of the Pb line similar to these of Cu indicate on uncovering of one of the primer layers. This is accompanied by an intensity increase of the Fe bands which indicate that the Fe-based primer layer lies closer to the copper sheet and was applied first. 4.5 Wax, primer and corrosion products

1464 1441 1418

1296

1061

1172 1133

The strong soiling and partial damage of the external, protective wax layer of the figure forced its complete removal for restoration purposes which was described earlier in the text. From the other hand, independently whether this layer was first applied during the previous conservation intervention (∼1950’s) or it represents the original old technology—its protective effect was positive and well proven. In order to identify this layer its chemical composition was analyzed by means of the micro-Raman technique. Due to the variety of waxes, their origin and differences in composition being a mixture of saturated compounds (esters, fatty acids, alcohols, hydrocarbons) (Vandenabeele et al. 2007, Edwards et al. 1997) the spectra were measured for the wax sample extracted from figure and also for the reference samples: beeswax, paraffin and microcrystalline waxes (including Cosmolloid H80). It was found that bands recorded in the spectral range between 1000 and 1550 cm−1 comprise the distinctive features and

comparison of the spectra is presented in Figure 9. The most prominent bands belong to: hydrocarbon bond CH2 (1464 cm−1 −bending vibration, scissoring; 1441 cm−1—bending, asymmetric; 1418 cm−1 and 1296 cm−1—bending, turning and twisting); carbon bond C-C (1172 cm−1, 1133 cm−1—stretching), and the stretching mode of the C-O bond centered at 1061 cm−1. Bands of the both reference crystalline waxes at 1418 cm−1 and 1464 cm−1 have similar intensities, while the investigated sample reveal intensity difference which discriminate these references. In case of paraffin the relative intensity of two strongest bands at 1296 cm−1 and 1441 cm−1 of I1296/I1441 = 1.3 is markedly different from the ratio of 0.6 observed for the historical sample, thus paraffin is eliminated, too. This indicates on the beeswax as originally applied protective layer and is confirmed by the strong similarity of both spectra. The Raman bands recorded for two another locations not covered with the wax layer indicate on the presence of antlerite, calcium carbonate and carbon. This is shown in Figure 10a, b together with the corresponding reference bands. These characteristic for antlerite are centered at: 267 cm−1, 416 cm−1, 484 cm−1, 602 cm−1, 990 cm−1, 1079, 1174 cm−1, and for carbon at: 1325 cm−1, 1580 cm−1, and coincide with the reference data—Figure 10a. The Cu3SO4(OH)4 is ascribed to corrosion of the Cu substrate, and carbon to atmospheric contamination (soot). The presence of CaCO3 (calcite) was confirmed for the pellet prepared of the grayish powder found in closed profiles of the figure and is ascribed to remnants of the extender which mixed together with oil and with Fe or Pb oxides was used for grounding. It cannot be excluded however, that

sample

beeswax micro crystalline 1 micro crystalline 2

paraffin 1000 1100 1200 1300 1400 1500 1600 1700

Raman shift / cm-1 Figure 9. Comparison of the μ-Raman spectra of the protective organic layer (sample) and reference materials.

165

LACONA8.indb 165

11/10/2010 5:37:12 PM

a)

50

40

sample 1 (St George - corrosion) antlerite Cu3SO4(OH)4 carbon

18

16

b)

is not clear. The clearly visible peak of zinc together with copper explains the gold-like appearance of this gilding made of the Cu-Zn alloy (“schlagmetall”). Moreover, it supports the conclusion on different primer materials and also gilding techniques applied originally and during previous conservation interventions.

sample 2 (St George - contamination) calcite CaCO3

Intensity [ a.u. ]

14 30 12 20 10 10

5

8

0

6 500

1000

1500

500

1000

1500

Raman shift [ cm-1]

Figure 10. The μ-Raman spectra: (a) of the corrosion products and surface soiling (solid line) with spectra of carbon (dashed line) and antlerite (dotted line) from the library, and (b) of the surface contamination (solid line) with reference spectrum of the calcite (dashed line).

CONCLUSIONS

In frames of the conservation project of the XVI c., gilded figure presenting St. George killing the dragon the documentation and results of the materials investigation of the object were obtained by means of equipment available at the Pomerania Laboratory at IF-FM in Gdańsk. The digital documentation of the object shape was made for reconstruction and virtual presentation. From coincidence of the results obtained by means of complementary use of the chemical (XRF, LIPS, EDS and m-Raman) and structural analysis (stratigraphy and SEM), the materials and techniques applied in the past have been concluded. Indications regarding the figure conservation were obtained from data on the thickness and elemental composition of the subsequent layers such as the protective ones (wax, gilding), primers, these of corrosion products and contaminants. Moreover, for removal of the soiled beeswax layer its detachment from the gilding by the pulsed laser radiation was successfully applied. REFERENCES

Figure 11. The EDS spectrum of the gold-free gilding (a) and the microstructure of the corresponding surface spot, magnification × 500 (b).

the amount of microcrystalline particles of calcite on the figure surface and especially in the closed profiles results from accumulation of dust over many years. This conclusion agrees in general with the data on elemental composition obtained from the XRF analysis. The EDS data obtained for the part of gilding where the gold layer is not observed confirm the presence of elements Ca, Cu, Fe, Pb on the inhomogeneous surface which reveal numerous pores and local material separations—Figure 11a,b. Elements Si and O may be ascribed to SiO2 contained in the surface soil similarly to calcite. However, the presence of Al may indicate content of clay or one of its components (e.g. kaolinite-Al2Si2O5(OH)4) in the primer. The appearance of lines ascribed to tin and phosphorus

Edwards, H.G.M. & Falk, M.J.P. 1997. Fouriertransform Raman spectroscopic study of unsaturated and saturated waxes. Spectrochimica Acta Part A 53: 2685–2694. Friedrich, J. 1997. Gdańskie zabytki architektury do końca XVIII w. Gdańsk: R. Ziarkiewicz Publ: 194–196 (in Polish). Oujja, M., Vila, A., Rebollar, E., García, J.F. & Castillejo, M. 2005. Identification of inks and structural characterization of contemporary artistic prints by laser-induced breakdown spectroscopy. Spectrochimica Acta Part B 60: 1140–1148. Sawczak, M., Kamińska, A., Rabczuk, G., Ferretti, M., Jendrzejewski, R. & Śliwiński, G. 2009. Complementary use of the Raman and XRF techniques for nondestructive analysis of historical paint layers. Applied Surface Science 255(10): 5542–5545. Vandenabeele, P., Ortega-Aviles, M., Castilleros, D.T. & Moens, L. 2007. Raman spectroscopic analysis of Mexican natural artists’ materials, Spectrochimica Acta Part A. 68: 1085–1088. Zheng, M. & Du, W. 2006. Phase behavior, conformations, thermodynamic properties, and molecular motion of multicomponent paraffin waxes: A Raman spectroscopy study. Vibrational Spectroscopy 40: 219–224.

166

LACONA8.indb 166

11/10/2010 5:37:12 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

THz-Time-Domain Spectroscopy—A new tool for the analysis of artwork M.J. Panzner, U. Klotzbach & E. Beyer Fraunhofer Institut für Werkstoff- und Strahltechnik, Dresden, Germany

G. Torosyan Fraunhofer Institut für Physikalische Messtechnik, Kaiserslautern, Germany

A. Schmid Hochschule für bildende Künste, Dresden, Germany

W. Köhler Labor Köhler, Potsdam, Germany

ABSTRACT: The analysis of artwork offers many examples, where the dream to look behind covers could reveal valuable results. We are currently engaged in experimental work using the THz-Time-Domain (TD) technology as a new tool that could realize this dream, because THz radiation can pass through many materials without substantial attenuation. Utilizing very short pulses in the THz-TDS (THz-Time-Domain Spectroscopy) allows tomographic investigations using runtime measurements. Simultaneously, they deliver a broad spectrum of electromagnetic waves with frequencies up to 5 THz for spectroscopic investigations. However, many experiments have yet to be carried out in order to establish this method for reliable artwork investigations in the future. Some of them are presented in this report. The visualization of hidden wall paintings requires some knowledge about the physical fundamentals, for example the scattering behavior of the granular structure of the wall material. Moreover, we looked for answers on questions concerning the detection of paint pigments by means of THz-TDS using the reflection mode setup. 1

INTRODUCTION

The THz frequency regime is roughly defined from 0.1 GHz to 15 THz. The wavelengths λ are accordingly within a range of 3 mm to 20 μm, thus this regime falls between the infrared and microwave regions of the electromagnetic spectrum. Therefore, the THz region spans the gap between electronics and optics. The special properties of THz radiation waves enabled them for a wide variety of applications, for example security scanning devices at airports. THz radiation passes through many non-polarized materials nearly without attenuation. Hence the detection of items through clothing and plastic bags is possible. However, polarized substances (like water vapor) show absorption lines in the THz-spectrum. Electromagnetic pulses with a pulse width of approximately one picosecond and a spectral width up to 4 THz (λ = 75 μm to 3 mm) were used when the Time Domain Spectroscopy (TDS)—technique was applied for analytical investigations. This method allows the tomographic

analysis of the sample structures, considering that THz pulse echoes can only be measured as long as refractive index jumps occur. The simultaneous calculation and conversion of the pulses from the time to the frequency domain by means of Fourier transformation result in a pattern of absorption lines, which provide the necessary information for unambiguous substance identification. One of the most recent applications of THztechnology includes the development of nondestructive analysis of artwork. Further potential applications are certainly conceivable by continuously improving the capabilities of THz- technology. The most exciting idea, namely to visualize hidden medieval wall paintings covered behind different materials like plaster or whitewash, was mentioned in 2006 (Köhler & Panzner et al. 2006). The first attempt to use THz-visualization to detect a graphite underdrawing below paint was published in 2008 (Jackson et al. 2008). Also, delamination processes or hidden cavities in sandstones can be located using the THz technology, as it has already been

167

LACONA8.indb 167

11/10/2010 5:37:13 PM

shown for ceramics on metal surfaces (Panzner et al. 2008/1). Furthermore, Panzner et al. 2007 demonstrated the possibility of non-destructive detection of pesticides by means of THz-TDS for Lindan and PCP. Latter application might be of particular interest for many museums, which preserve partially contaminated artwork made of textile or wooden materials. Fukanaga et al. 2007 reported about the spectroscopic identification of paint pigments by using THz-TDS-measurements in transmission mode. In the meantime, much data was recorded and subsequently implemented into a database, which nowadays contains a large number of pigment THz-spectra, which are easily accessible over the World Wide Web. However, below a frequency of 5 THz only few pigments appear as sharp lines in the spectrum to allow one an easy identification (Panzner et al. 2008/2). Of particular relevance for this report was to prepare and conduct some preliminary experiments on wall paintings by means of THz-TDS. Our work focused mainly on the transmission issue of wall painting covers, scattering effects of THz-radiation on granular wall materials as well as the question, whether utilizing an experimental setup in reflection mode yields any detection of pigments at all.

2

EXPERIMENTAL

2.1

Experimental setup

Our results, which are discussed here, were measured using different experimental setups. Laboratory setups of the Fh IPM where used for both transmission and reflection mode. Reflection measurements were carried out using an incident angle of zero degree, because it eliminated the polarization dependence of the THz-TDS results. The main components of a THz-TDS setup are shown and extensively explained in Panzner et al. 2007. A titanium-sapphire-femto second laser with a lock-in amplifier synchronized to a mechanical chopper wheel was used for our laboratory work. THz-pulse emission was induced by laser shots onto a photoconductor by using the effect that an electrical field caused by charge dipole is formed in the vicinity of a semiconductor surface after light generated charge carriers (photo-Dember effect) (Zhang X.-C. et al. 1990). An Auston switch (Auston et al. 1984) was used as a detector for all experimental setups. Preliminary transmission experiments on plaster samples were carried out using a fiber coupled system (Ellrich et al. 2008) to demonstrate the capability of such a system, which will eventually be used to discover hidden features in artwork by means of reflection measurements.

Figure 1.

Cross section of the plaster sample.

2.2 Samples Quartz powder, prepared by the company KREMER PIGMENTE, was used for the THz scattering experiments. Different quartz powders with grain sizes of: 0,04–0,15 mm; 0,1–0,25 mm; 0,25–0,4 mm; 0,4–0,5 mm and 0,5–1 mm were supplied. The different grain size groups were obtained using a grading sieve. The powders were filled into cuvettes (12 × 12 × 35 mm3, wall thickness 1 mm) so that a propagation path of 10 mm had to be passed by the THz-pulses for both transmission and reflection experiments. As almost no attenuation occurs during the transmission of the THz radiation, the refraction index difference to that of air causes the pulses to get reflected by each interface layer nevertheless (Fig. 4a). A plaster sample (thickness 12 mm) as it can be found behind hidden wall paintings was prepared for further transmission and reflection experiments. Fig. 1 shows a cross section of this sample. We used three samples that contained the pigment cinnabar to look for signatures typical for a pigment in transmission as well as in reflection experiments: − cinnabar pigment powder filled in a cuvette with a transmission length of 4,3 mm − pressed tablets of a mixture cinnabar and HDPE − typical paint removed from a smooth surface consisting of cinnabar in a binder rabbit-skin glue. Latter skin-like-sample with the thickness of about 30 μm was mounted in a special plastic frame. 3 3.1

RESULTS Transmission and reflection of THz waves on plaster

A plaster sample with a thickness of 12 mm was measured using THz-transmission to demonstrate the ability of a fiber based THz-system to provide enough THz-power to investigate hidden structures

168

LACONA8.indb 168

11/10/2010 5:37:13 PM

like wall paintings or cavities. Fig. 2 shows the recorded transmission spectrum. The plot shows transmission of about 20–30% occurs at approximately 0.2 THz, whereas for higher frequencies the transmission decreases rapidly. 3.2

THz-wave scattering at quartz powders

The attenuation of THz radiation through plaster is less caused by material absorptions, but mainly by scattering at solid sand grains. This can easily be demonstrated by TDS- transmission and reflection measurements using quartz powders with different grain sizes. Fig. 3 shows the spectrum of numerous quartz powders, which were transformed from the time transition domain signals applying Fourier transformation. Only the low frequencies with wavelengths significantly longer compared to the grain size were transmitted with attenuation. Small grain sizes don’t influence the transmission of long wavelengths as much as large grain sizes. The strongest

Figure 4. STFT calculated from the time domain signal which results from reflection (incident angle 0°) at a) an empty cuvette, b) an cuvette filled with quartz powder (grain size: 0,5–1 mm).

Figure 2. THz -Transmission spectrum of the plaster sample (Fig. 1).

Figure 3. Transmission spectrum of quartz powders with different grain sizes.

scattering occurred for grain sizes of 0.5–1 mm, which matched the utilized THz-wavelengths. The scattering behavior can be understood easily by looking at the responsible mechanisms. If grain size diameters approach THz-wavelength dimensions, strong Mie scattering occurs, whereas for smaller grain sizes Rayleigh scattering becomes the predominant mechanism (Paetzold H.K. 1952). Likewise, scattering effects can be observed working in reflection mode. Short Time Fourier Transformation (STFT) is a suitable tool to investigate which part of the spectrum is reflected in which depth. Fig. 4 shows the STFT results of THZ-TDS- reflection at an empty cuvette and another one filled with quartz powder. The STFT-graphs show which frequencies arrive the detector over time. The subsequent analysis uses the same principle already well-known from radar depth measurements, which means that the time scale provides information about the penetrated depth of the signal that is reflected. With the sample chamber empty, back reflections occurred at all interfaces of the cuvette. Although no back reflections could be detected from the back side of the quartz powder filled

169

LACONA8.indb 169

11/10/2010 5:37:13 PM

chamber, a large portion of the beam penetrated into the quartz powder and was then backscattered. The values of backscattered frequencies from deeper regions are lower than those reflected directly from the surface. 3.3

Pigment identification in transmission and reflection

Only some pigments show signatures in the wavelength regime of THz-TDS. The transmission results of three of them are shown in Fig. 5. Here the challenge was to identify these pigments in reflection mode. If overcoming some obstacles that still exist, it could be the first step to successfully realize our vision to visualize hidden wall paintings. And since the pigment cinnabar exhibits strong absorption at 1.2 THz, it was used to investigate its reflection characteristics. For an incident angle of 0° the reflection can be shown to be of the form (1), which is derived from Fresnel’s equation: ⎡ n − 1⎤ R=⎢ ⎥ ⎣ n + 1⎦

Figure 6. Absorption coefficient and refraction index of a pressed tablet of 50 mass−% cinnabar in HDPE.

2

(1)

where n is the refraction index of the material in the case of reflection in dry air. Fig. 6 shows the refraction index and the absorption coefficient of cinnabar calculated from THz-TDS-measurements at the pigment powder. Typical absorption lines for cinnabar occurred near 1.2 THz. Moreover, clear anomalies even appear in the detected resonance frequencies at the refraction index graph (light grey graph, Fig. 6). Because of (1) a similar behaviour should be found for the reflection coefficient. Fig. 7 shows the degree of transmission of pigment powder, in form of a pressed tablet, containing 16% cinnabar and a skin of cinnabar in a binder of rabbit-skin glue

Figure 7. Transmission and reflection behaviour of different cinnabar compounds (paint skin—cinnabar in rabbit skin glue, Tablet—16.7 mass−% cinnabar in HDPE, pigment powder—cinnabar in a cuvette, transmission length 4.3 mm).

typically used in paintings. Additional measurements on this skin in reflection mode (light grey graph) are included. Clearly, the expected step at the resonance frequency can be seen at frequencies of 1.17 THz and 2.66 THz, where all materials show signatures (absorption minima) in transmission. Because the pure pigment powder shows strong attenuation at 2.6 THz, the according signature virtually disappears due to too much noise. 4

CONCLUSIONS

Applying electromagnetic waves in the THz frequency regime for the investigation of artwork can help to discover a large number of additional results, for example determining layer thicknesses or grain sizes of materials. The results obtained by THz-TDS measurements are strongly influenced by the scattering of THz-waves caused by

Figure 5. Transmission of three paint pigments showing resonances in the THz-TDS-range.

170

LACONA8.indb 170

11/10/2010 5:37:14 PM

the natural grain-like structure of the investigated materials. Furthermore, wall paintings showed additional signatures in the spectrum, caused by the varying water content or its layered structure. These complex correlations make an easy interpretation of obtained results difficult. Characteristic signatures of a pigment could be detected by conducting measurements in reflection mode. We believe that this is the first step on the road to realize the vision of visualizing hidden wall paintings. The investigations presented in this report can help to evaluate future results obtained from measurements on hidden wall paintings. ACKNOWLEDGEMENT This work was part of the project TERAART (Grant No.: 01UA0805 A), which was funded by the BmbF (Federal Ministry of Education and Research) and reviewed and supported by the DLR (German Aerospace Center). We gratefully acknowledge the support.

Fukanaga, et al. 2007. THz spectroscopy for art conservation. IEICE Electronic Express 4 (8): 258–263. Jackson, J.B. et al. 2008. Terahertz imaging for nondestructive evaluation for mural paintings. Optics Communications 281: 527–532. Köhler, W. & Panzner, M. et al. 2006. Non destructive Investigation of Paintings with THz-Radiation. ECNDT Berlin 2006, Proc. 9th Int. Conference on NDT, Berlin, 25–29 September 2006. Paetzold, H.K. 1952. Ein Beitrag zur atmosphärischen Extinktion. Astronomische Nachrichten 281: 17ff. Panzner, M. et al. 2008/1. Investigation of hidden Structures in Art and Technology by THz-Time-Domain Imaging. 22nd International Conference on Surface Modification Technologies, Trollhättan, 22–24 September 2008. Panzner, M. et al. 2007. Potential of THz-Time-Domain Spectroscopy in object inspection for restoration. Lasers in the Conservation of Artworks – Castillejo et al. (eds) 2008 Taylor & Francis Group, London, ISBN 978-0-415-47596-9. Panzner, M. et al. 2008/2. Application of THz-Technology for the Investigation of Art Objects. 3rd Workshop on Terahertz Technology, Kaiserslautern, 4–5 March. Zhang, X.-C. et al. 1990. Optically induced electromagnetic radiation from semiconductor surfaces. Appl. Phys. Lett. 56/22: 2228–2230.

REFERENCES Auston, D.H. et al. 1984. Picosecond photoconducting Hertzian dipoles. Appl. Phys. Lett. 45: 284–286. Ellrich, F. et al. 2008. Fasergekoppeltes TerahertzSpektroskopiesystem. Technisches Messen 75 (1): 14–22.

171

LACONA8.indb 171

11/10/2010 5:37:21 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

19th century paints of Richard Ainè used by Jan Matejko (1838–1983). Analysis of preserved paints from tubes, palettes and of paintings’ surfaces and paint-layer M. Wachowiak Department of Conservation and Restoration of Modern Art, Nicolas Copernicus University, Torun, Poland

ABSTRACT: Subject of the research are paints used by 19th century Polish academic painter Jan Matejko (1838–1893). Pigments’ elemental composition of colours—from preserved tubes and from two palettes from Krakow collection, as well as from his paintings from 1880-ies—was examined by means of XRF and Raman spectroscopy. Research proved importance of preliminary study of tubes and palettes in order to identify later pigments on paintings only basing on non-invasive methods. Model spectra of pure samples were crucial in paintings analyses, especially in the case of numerous ready made market mixtures of pigments. Copper and arsine containing greens; cadmium, strontium and chrome based mixtures of yellows as well as mixtures used for body flesh tones and numerous diverse kinds of modified Naples yellows were identified in tubes, palettes and on paintings. Also carriers and fillers of organic dyes let to distinguish organic reds and yellows in tubes, on palettes and on paintings. Some characteristic features of artist technique were established like wide range of yellows used (especially of Naples yellows), iron reds of two kinds (rouge de Venise and Rouge de Pozzuoli) and Van Dyck Brown used widely in underpaintings and shadows. Unexpected there was hardly any vermillion or iron—based dark browns. Numerous analogies in elemental composition between paints from tubes collection and palettes and pigments on paintings were observed. Cobalt green first used in 1883 was established as dating factor. 1

INTRODUCTION

Collection of paints and of palettes of Jan Matejko preserved in Krakow collection as well as some of his paintings are the subject of the preliminary research concentrated on elemental composition of pigments. Tubes of oil paint of Jan Matejko from the Richard Ainè company from Paris and some powder pigments were analysed. Two of the seven preserved palettes can be prescribed to the paintings executed. One palette has original, painted inscription of the artists, informing that it was used to paint Joanna d’Arc finished in 1886. Another palette has signature and date 1888scratched into its surface. In this year only one big painting The Raclawice Battle was finished. These two palettes and paintings were expected to show analogies in the character of pigments used. Some further examination of the palettes and paintings were done to broaden and confirm gathered observations.

using a portable spectrometer ArtTAX (Bruker). Excitation was provided by a Rh target x-ray tube. The measurement head with electro-thermally cooled silicon drift detector combined with a colour CCD camera were fixed on a tripod with motor driven XYZ stage for sample positioning. The measuring conditions were: voltage 50 kV, intensity 700 mA and live time 150 s. Some samples were identified by means of the confocal micro-Raman spectrometer (InVia, Renishaw) with the diode laser emitting at 785 nm. The spectral footprint of the sample surface area was 5 mm × 5 mm while magnification of 50 x. Spectra were acquired among 100–3200 cm_1 with resolution of 2 cm_1. Depending on the signal intensity up to 20 spectra were accumulated. The total acquisition time did not exceeded 5 min. 3 3.1

2

EXPERIMENT

The elemental analysis was carried out by micro energy dispersive X-ray fluorescence spectrometry (μ-XRF). Measurements were performed

RESULTS Tubes

From 273 tubes of paint of Richard Ainè in the Krakow collection, 30% percent are yellows, another 30% iron based ochres and reds. Surprisingly the smallest group, are iron-free reds, of number smaller than 5% of all colours.

173

LACONA8.indb 173

11/10/2010 5:37:21 PM

The group of yellows, being the biggest, shows also richest variety of kinds. There are five different hues of Naples yellow. Main difference between them is number of counts of zinc in ratio to lead and antimony (Tab. 1.). Another pigment identified is pure strontium yellow. Yet there appears also original producer mixture of strontium yellow with cadmium yellow and of chrome yellow with cadmium yellow of intensive orange hue. Organic yellows in the collection are—laque de Robert (N°5) and of darker hue laque de Robert (N°6), as well as still de graine, jaune de Indien and another orange hue organic yellow—probably jaune de garance. In group of powder pigments also massicot, orpiment and another organic yellow and Naples yellow are present. Ready made pinkish body flesh tone paints called jaune de brillant, are based on lead white modified by addition of little amount of chrome yellow, and, when in darker hues—also of cadmium yellow and vermillion. Big group of iron—based pigments, apart of iron itself, have characteristic elements, allowing to identify them on the palettes but when mixed on the paintings, it is difficult to distinguish them. Yet, few from the group of red earths which are distinguishable is strontium and barium containing rouge de Venise and arsine containing rouge de Pozzoulli. Unexpected there is only one small tube of oil paint vermillion present in the collection. Another red in powder pigment is lead red and two other hues of vermillion. Few organic reds of various tones characterize different fillers like chalk, and carriers for the dye, as aluminium or tin compounds. Blues represent cerulean blue, two different kinds of cobalt blue as well as three ultramarines and one tin and copper containing blue. There are three different copper—based greens. Two of them contain copper and arsine of significantly different ratio of counts to each other. One of two chrome based greens is viridian and the other presumably viridian with some addition of cobalt blue.

Table 1. Different ratio of counts for Sb, Zn, Pb in different kinds of Richard Ainè Naples yellow from the Krakow collection. Ratio of counts Colour name

Sb

Zn

Pb

Jaune de Naples claire Jaune de Pinart Nr 2 Jaune de Naple Jaune de Pinart Nr 3 Jaune de Pinart Nr 1

1 1 1,5 3 18

11 1 1 1 1

6 9 20 62 181

Unexpected among numerous iron—based colours there were any dark browns like burnt umber. Main brown—used by artist especially in the underpainting stage was the van Dyck brown. 3.2

Palettes

After collecting data for tubes, identification of paints samples of the palettes was done. Examining palettes, it has to be remembered, that it didn’t have to be used only for one painting, and while creating a picture, more than one could have been used. One palette has less colours and most of them of warm hue without any green. It is likely to be used only for painting body completion tones. Other palettes represent full range of colours. There is repeatable order of paints layout in all palettes: warm yellows lay at the right side, then reds, finally colder and darker blues and greens situated on the left lower part. There are pigments identified from samples from tube which were not found on palettes and some are present on palettes but not in tubes. Yet presence of huge number of colours, among them light yellows and iron based ochres and reds, is parallel. 3.3

Palette 4

On palette nr 4, prescribed to the painting Joanna d’Arc, numerous kinds of Naples yellow analogical to ones in tubes are present. The ready-made market mixture of strontium and cadmium yellow and orange market mixture of cadmium and vermillion which was absent in tubes, has been identified as well. Paint present only on the palette 4. is zinc based, lemon hue yellow, probably zinc chromate yellow, with small addition of cadmium yellow. Organic yellows of palette 4. are laque de Robert, jaune de Indien and still de graine. There were body flesh colours on both of the palettes repeating the same elemental composition as in tubes, as well as slightly different ones, with addition of iron, but close in hue. Reds identified on palette 4. are iron rouge de pozzuoli and organic reds of different hues, precipitated on alum or on tin based compounds. Carriers and fillers like chalk or some trace elements as potassium (probably coming from potash used in process of production), as well as phosphorus and silica let to distinguish individual kinds of organic reds. Other characteristic elements are arsine, copper and in one case also small amount of cobalt. Blues of palette 4 are cerulean blue and cobalt blues with small additions of elements, different from blues which come from tubes. Three kinds of ultramarine present in tubes close to each other in their elemental composition were impossible to distinguish on palettes with XRF.

174

LACONA8.indb 174

11/10/2010 5:37:21 PM

painting is rouge de Pozzuoli—iron red containing arsine. Joanna d’Arc is the only Matejko painting till now—on which cerulean, among cobalt blue and ultramarine—was identified. Number of greens is much smaller then the 7 kinds found on palette 4.: only copper green (probably malachite) and viridian were identified in the taken samples. But research will be still continued with portable spectrometer when possible after restretching the painting which is now being conserved. Then, when data will be broadened, it will presumably show more analogies in paints elemental composition for palette 4 and painting Joanna d’Arc.

Table 2. Different ratio of counts for Cu and As in different kinds of Richard Ainè light greens from the Krakow collection. Ratio of counts Colour name

Cu

As

Green 1 (Scheele’s green) Green 2 (Emerald green) Green 3 (Emerald green)

1 2 4

2 1 1

Copper based green present on palette 4. is probably malachite (or close synthetic version of this colour). Greens containing both arsine and copper can be divided into three groups for the reason of different ratio of these elements. The third with significantly smaller amount of arsine in ratio to copper was not present in tubes. All three kinds of greens are present on palette 4., (Tab. 2.) as well as typical viridian. Another two greens identified are: green earth and cobalt green with big amount of zinc, so called Rinmann’s green. Dark colours are Brun van Dyck, carbon black and bone black. 3.4 Palette 6 On palette 6. apart of other yellows also masssicot is present. Organic yellows represents only still de graine. Iron rouge de Venise on the palette 6. is used instead of similar in hue rouge de Pozzuoli. One distinct kind of cobalt blue on palette 6. contains apart of cobalt also molybdenum. There is Prussian blue too. Emerald greens represent the two kinds with smaller number of counts for arsine then for copper. The rest of colours on palette 6. repeat these identified on palette 4. On both palettes nor lead red nor orpiment was found, present in collection as powder pigments. 3.5 Paintings Elemental composition of paints from both palettes were compared to prescribed paintings, to which they were used while painting: palette 4. to the Joanna d’Arc painting and palette 6. to the Racławice Battle. 3.6

Joanna d’Arc

From the Joanna d’Arc painting 10 samples of paint-layer were taken. Colours identified repeat richness of yellows (Naples yellows, strontiumcadmium yellow, cadmium yellow, cadmium deep orange as well as massicot). Characteristic iron red which appears both on palette 4. and on the

3.7

Racławice Battle

Painting The Racławice Battle (1888) shown a lot of similarities and some differences to the palette 6. Pure strontium yellow and massicot not found on the palette were identified on the painting surface. Iron based red Rouge de Venise—was identified both on the painting and on the palette. Two blues not present as tubes of paints but only on the palette 6. were identified on the painting as well: copper based blue (azurite?), and in some parts Prussian blue. Another pigment identified only on palette 6. and The Racławice Battle painting is green containing both chrome and cobalt. There were any emerald greens found on the paintings contrary to the ones on the palette. Measurements on some other paintings from 1880-ies show similarities to the preliminary studies on model samples of pure paint from tubes and from palettes. Wide range and often use of different kinds of analogical Naples yellows was observed, some of them also slightly different, containing very often small amount of iron. Characteristic minimal amount of vermillion present in the collection was confirmed on paintings too, even on ones with huge areas of red colour. Main red is iron red rouge de Pozzuolli, often mixed with organic red. Another red paint on the painting (lacking on the palette but present as powder pigment) is lead red. Among other greens there appear cobalt green. Its occurrence on painting is the first time noted on painting in the sketch of Joanna d’Arc from the year 1883. In the same year it appears at Wernyhora and later also in Portrait of Zyblikiewicz (1887). 1883 is also the year when cobalt green is the first time mentioned on the shopping list of paintings materials bought by Matejko, since occurrence of cobalt green can be treated as a dating factor. Organic colours especially yellows are hardly identified on paintings. Their occurrence is very probable only in few cases. Organic reds were established but still the identification of these colours is difficult, till amount of filler like chalk or carriers as tin is not significant enough. Yet, arsines based organic red,

175

LACONA8.indb 175

11/10/2010 5:37:22 PM

as well as ones with tin containing carrier or with cobalt trace addition are distinguishable. More precise identification needs complementary methods like Raman spectroscopy and FTIR. The same is in the case of blues, really hard in interpretation, apart of cerulean and cobalt blue. Ultramarine is problematic, moreover that it appears normally with strong addition of lead white. 4

CONCLUSIONS

Comparing data from paintings with model samples from tubes and from palettes proved to be very important. There was no zinc white identified in Matejko’s paintings yet zinc appears often in lead and antimony based Naples yellows. Without knowledge that it is ready made market composition of yellow it could led to wrong conclusion that it is an artist addition of zinc white to the yellow colour. Also in the case of different kinds of copper and arsine containing greens possibility to distinguish them on the base of ratio of counts for main elements was shown. Numerous pigments appeared to be ready made producer mixtures as in the case of cadmium yellows and strontium and chrome yellow. Also very characteristic pink flesh colours found on artists’ paintings, appeared to be ready made producer colour and can be treated as well as rich variety of different Naples yellows, as one of the individual features of the artist technique. Another distinct feature is hard use of vermillion and common use of iron reds rouge de Pozzuoli or rouge de Venise instead of other reds. Significantly Brun van Dyck replaced burnt umber and other iron—based browns. Palettes contained most of the pigments identified in tubes. Few pigments appeared only in tubes, some only on palettes. It is likely that there were more palettes used to execute the two analysed paintings then palette 4. and palette 6. Powder pigments: massicot, Prussian blue, red lead—absent on palettes but present on paintings—were presumably grinded independently of palettes. Occurrence of

some paints on palette and their lack in tubes, as well as some names of paints from the shopping list of paintings materials bought by Matejko which were unable to connect with names from labels on preserved tubes, suggest, that number of kinds colours in tubes was slightly bigger. Research of tubes, palettes and paintings of one Matejko show numerous connections among them and great analytical capabilities strengthening each other when data is compared. The research, being at the preliminary stage will be continued with wider use of Raman spectroscopy as well as FTIR and EDX to examine deeply chemical composition of all paints.

ACKNOWLEDGEMENTS Research was supported by stipend “Step forward—second edition of stipends for doctorals”, (European Social Founds, Polish Government, province of Kujawsko—Pomorskie). Special thanks are to: Bruker company for non-paid loan of XRF portable spectrometer in the year 2008, dr Anna Klisińska–Kopacz from LANBOZ laboratory in Krakow and to dr M. Sawczak from Polish Academy of Sciences, The Szewalski Institute of Fluid-Flow Machinery, Gdansk.

REFERENCES Biasion, 1883? unpublished handwritten source - selling book from Biasion shop of painting materials in Krakow, notes for years 1880–1883, National Museum in Krakow (in Polish). Carlyle, L. Authenticity and adulteration: What materials were 19th century artists really using?, The Conservator, 1993, 17, 56–60. Eastaugh, N., Walsh, V., Siddal, R. & Chaplin, T., 2004, Pigment Compendium - A dictionary of historical Pigments, Oxford, Elsevier. Hopliński, J. Farby i spoiwa malarskie, 1995, Wrocław, Ossolineum.

176

LACONA8.indb 176

11/10/2010 5:37:22 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Study of the effect of relative humidity on the identification conditions of paper soiling by means of the NIR technique M. Sawczak, G. Rabczuk, A. Kamińska & G. Śliwiński IFFM Polish Academy of Sciences, Gdańsk, Poland

ABSTRACT: The work is focused on the impact of changes of the Relative Humidity (RH) on the NIR spectra of selected model samples of contemporary paper, which are soiled with substances considered as staining as well as those considered as safe and used in paper conservation, too. The effect of fluctuations of RH during the spectra measurement resulting in changes of the water content of paper samples is analyzed and correlation with the spectra is discussed. With an increase of RH a linear increase in the intensity of spectral response is observed in the regions corresponding to the main absorption bands of water in the NIR (1350–1520 nm and 1850–2000 nm). For a given, soiled paper sample, the sensitivity of the spectral response to RH changes measured by the slope of the regression line showing the dependence of the peak intensity of the band 1850–2000 nm on the RH, characterizes the soiling applied. This feature can be applied as an additional indicator for the samples discrimination. The analysis of the discrimination conditions of samples showing spectral features characteristics for the soiling substances is performed by means of the PCA technique. Results show that the influence of relative humidity changes on the samples discrimination can be considerably reduced by analyzing the second derivative of the spectra. Spectra analyzed with masking of regions of the water absorption bands confirm significant improvements in discrimination of the soiled paper samples for the soiling substances of characteristic absorption bands not located in these regions. In contrary, for substances revealing band overlap with these of water the masking procedure significantly reduce the possibility of the proper spectra classification. 1

INTRODUCTION

The scientific approach to the conservation practice leads to better understanding of the artefacts nature and the way of their deterioration. The identification techniques are helpful while establishing the most appropriate restoration methods. Discoloration, soiling or stains are one of the most common examples of paper decay factors. The exact and sensitive examination technique is able to provide conservators with essential information for example: identification of the staining factors enables the conservators to choose the optimal method for removing stains. On the other hand when the factors are harmless to the paper and there is no other contraindications like illegibility of the image or the writing, the examination results may suggest abandoning the cleaning process in consequence. A number of techniques has been employed in the paper conservation for identification of the deterioration factors. Most often they require some amount of the sample that is not possible in some circumstance. However, the most reliable identification techniques relay on the complex, expensive and time consuming methods like GC, GC-MS, TLC, HPLC, etc. There is

a great need for an inexpensive and non-invasive technique that is capable of providing the accurate and reliable examination. Near Infrared Spectroscopy is one of the techniques widely used in the pharmaceutical, chemical or in the food industry for the identity testing of raw material and finished products (Guo-quan 2006, Candolfi 1997 & Blanco 2008). This method is often employed for rapid evaluation of paper conditions (Ali 2001 & Yonenobu 2009). The wavelengths from the NIR region between 900 and 2500 nm are characterized by a weak absorption. Thus, a large part of the incident light is reflected from the solid surface of the sample and can be collected and analyzed in the reflectance mode. Various chemometric methods are used for the analysis of the resulting spectral characteristics (Næs 2002). Results presented in this work are obtained in frames of the research project aimed on application of the NIR spectroscopy for identification of substances present in the form of different contaminations or staining on paper. For the purpose of this study the contemporary model samples are applied. These results from the limited access to historical paper samples, and the necessity of

177

LACONA8.indb 177

11/10/2010 5:37:22 PM

their special treatment justify the choice of the modern paper samples at the present stage of the project, too. Moreover, the usage of the contemporary material enables to evaluate the capability of the NIR technique to identify the stains on different types of paper which can be applied as valuable reference data in analysis of the historical materials. 2

EXPERIMENTAL

The measurements of the humidity effect on the spectral characteristics of paper were carried out for two cases: − clean model paper samples − model paper samples coated with selected adhesives, coatings or substances commonly recognized as staining factors—Table 1. For preparation of the model samples the contemporary paper commonly used for printing, photocopying, drawings and newspapers has been Table 1. Chemical composition of model samples and substances applied as paper additives (soiling). Model papers

Composition

Paper 1

Drawing paper (60% eucalyptus, 36% pine, 4% cotton) Outprint paper (bleached sulphite softwood and groundwood cellulose pulp: 69% birch, 31% pine content) Whatman filter paper (cotton linters; 98% α- cellulose content) Newsprint paper (87% thermomechanical pulp, 8% sulphite groundwood (pine), 5% sulphite softwood cellulose pulp)

Paper 2

Paper 3

Paper 4

Paper additives (soiling) Subst1 Subst2 Subst3

Subst4 Subst5

Subst6 Subst7 Subst8

Composition Granulated white sugar Methylcellulose Paraloid B72 (acrylic resin: copolymer of methyl acrylate and ethyl methacrylate) Gelatin Butaprene (chlorobutadiene rubber, benzene, toluene, gasoline, other hydrocarbons) Definol WN (wheat starch) Linseed oil Vinavil Blue NPC (polyvinyl chloride and polyacetate water dispersion)

applied, and the Whatman filter paper was used as reference. The wheat starch and gelatin used as sizing since the beginning of paper production were selected for investigation because of the known sensitivity to the environmental changes, especially to the RH fluctuations. Similarly, the methylcellulose is commonly used for paper sizing, and also as paper adhesive easily washable with water. Also the glue Butaprene was choosen due to the brownish, hardly removable staining and serious damages caused to paper artefacts and documents which is observed by the paper conservators. This glue was used previously to fix the book or document covers and binding. In order to model soiling due to the human activity the sugar and oil were selected. Numerous types of soiling on paper can be caused by substances containing sugar in the form of monosaccharides (fructose, glucose of fruits, honey, etc.), disaccharides (e.g. saccharose added to drinks, sweets) or as an additive to water colour paints, etc. Similarly, the oil based substances and fats cause various greasy stains on paper. The soiled model samples were prepared for analysis by immersion of the pure paper samples in the 2% water solution of the substances listed in Table 1: gelatin, methylocellulose, Definol WN, Vinavil Blue NPC, saccharose; or 2% Paraloid B72 solution in acetone; 2% Butapren solution in toluene, or saturated with linseed oil, respectively. Next, the samples were dried at room temperature for 48 hours. The relative concentration of the substance applied to the paper was estimated from the sample weight comparison before and after preparation. During measurements of the NIR spectra the samples were placed in a climatic chamber with Relative Humidity (RH) controlled in the range of 20–80%. Spectra were collected in the diffuse reflectance mode using a custom-built scanning spectrometer equipped with a broadband radiation source (quartz tungsten bulb, 3000 K) and PbS detector. To avoid increase of the samples temperature during measurements resulting in its humidity change, the spectrometer was operated in the pre-dispersive mode, i.e. the source radiation was dispersed by the monochromator grating and directed to the sample chamber. This provided the sample irradiation at minimal intensity corresponding to the actually scanned wavelength. Spectra were recorded in the wavelength range between 1300 and 2500 nm with sampling interval of 2 nm and the data were converted to absorbance units (log(1/R)). The reference spectra were recorded before each sample measurement by scanning the reflectance standard (Spectralon).

178

LACONA8.indb 178

11/10/2010 5:37:22 PM

In order to minimize the radiation scattering effect due to inhomogeneity of the paper surface, five independent spectra were measured at different locations across the surface of each sample and the results were averaged. 3 3.1

RESULTS AND DISCUSSION Model papers

Examples of the NIR spectra measured for paper stored in air under conditions of variable RH (h) are presented in Figure 1. Main changes in the spectral characteristics vs h can be observed in two bands (1350–1520 nm and 1850–2000 nm) corresponding to major absorption bands of H2O (Workman 2007). On Figure 1b the differential data obtained by subtraction of the reference spectrum measured at hmin = 0.3 from data obtained for the relative humidity h changing in the range form 0.35 to 0.8 are shown together with the absorption spectrum of water. Beside the mentioned bands, the slight influence of humidity can be observed in the wide band above 2300 nm. The strongest water absorption band with maximum at 1940 nm is located close to the characteristic cellulose band with maximum at 2100 nm—Figure 1a. The second

Figure 1. NIR spectra of paper sample 3 measured for different values of relative humidity (a) and differential spectra of the same sample related to spectrum obtained for hmin = 0.3; the water absorption spectrum is shown for comparison (b).

one (1350–1520 nm) overlaps the cellulose band located at 1490 nm. Spectral regions 1340–1400 nm and 1800–1900 nm correspond to the weak bands of water vapor. The dependence of the humidity of paper vs. air represents an individual function of the paper properties (composition, physical conditions, etc.). This is shown in Figure 2 for the investigated samples as the dependence of intensity of the main absorption band of water (1850–2000 nm) on the air humidity in the storage chamber. While the water band intensity measured for the same h value depends on the sample, the dependences are linear functions of the relative humidity level. The slope of the characteristics is almost independent on the type of paper. Spectra measured for the 4 model papers (in total—44 spectra measured at eleven levels of relative humidity), were analyzed using discrimination techniques, in order to test the possibility of differentiation of various papers due to their composition, but independently from the variable RH of air during measurement. The different methods of data standardization were tested (MC, SNV, MSC, first and second derivative) to optimize the discrimination procedure. The PCA applied to the spectra separates them into four groups. Those groups are composed by spectra of papers characterized by a different composition—Figure 3a. The samples are distributed along the vertical axis within each group, according to the level of the relative humidity during the measurements. As a consequence groups of different papers overlap partially. In order to eliminate the influence of the paper humidity on the data clustering of different types of papers, the second derivative of the spectral data was analyzed. The PCA applied to the second derivative separated the spectra into four well defined groups and the influence of relative

Figure 2. Changes in intensity of the water absorption band at 1940 nm measured for four model papers stored in the air under variable RH conditions.

179

LACONA8.indb 179

11/10/2010 5:37:22 PM

It follows from data presented on Figure 4 and in Table 2, that the presence of substances like Vinavil, methylcellulose, and saccharose causes the largest changes of the water absorbance of paper while the influence of the Definol or Paraloid B72 seems to be moderate only. The above results are confirmed by a direct measurement of the samples humidity via observation of the sample mass change during its drying. Samples were weighted at room temperature and relative humidity of 0.4 and then heated up to 120°C. The drying process was finished when the sample weight stabilized and remained unchanged for 120 s. Data obtained from this RH measurement and the intensity changes of the water absorption band in the NIR spectra measured for the corresponding relative air humidity (h values) are presented on Figure 5. The data reveal, that changes of the water absorption band intensity (Imax) are similar in character to these of the samples humidity determined from direct measurements. This regularity is disturbed in case of the

Figure 3. First two principal components calculated for spectra of four model papers (a) and for second derivative of the spectra (b) measured at 11 levels of relative humidity changing form 0.3 to 0.8.

humidity within each group was significantly reduced making possible sample identification independently on the relative humidity changes during measurements—Figure 3b. 3.2

Model paper with soiling

The influence of relative humidity on the spectral characteristics of paper soiled by the 8 different substances listed in Table 1 was investigated for the Whatman paper under conditions identical as those applied for the samples without soiling. The corresponding intensity dependence of the water absorption band measured at 1940 nm vs. relative air humidity Imax(h) is presented on Figure 4. For all samples, similarly to the non-soiled ones, the linear dependences are characteristic. However, the difference in slopes confirms, that the water absorbance level of paper varies depending on the type of soiling/sizing/adhesive applied. This effect becomes more evident for the slope values related to that of the Whatman paper—see Table 2. The observed effect can be used as an additional indicator for discrimination of the contaminated/soiled papers and to differentiate the type of soiling, too.

Figure 4. The intensity changes of the main water band with maximum at 1940 nm measured vs. relative air humidity measured for Whatman paper covered different substances. Table 2. Slope of the characteristics Imax(h) measured for stained paper samples and related to the slope obtained for the Whatman paper. Paper soiling

Rel. slope of characteristic

Subst1 Subst2 Subst3 Subst4 Subst5 Subst6 Subst7 Subst8

1.29 0.32 0.05 −0.3 0.12 0.05 0.1 0.53

180

LACONA8.indb 180

11/10/2010 5:37:23 PM

Figure 5. Relative humidity of Whatman paper covered with different substances determined by direct measurement in relation to intensity of the band at 1940 nm, measured at air humidity of 0.4.

sample contaminated with linseed oil and can be explained by the error of direct humidity determination of that sample during its drying process, because some of the oil fractions may evaporate from sample which in turn lead to the change of mass. 3.2.1 Discrimination analysis The influence of the RH changes on the identification possibility of different substances applied to the model paper was tested by means of the discrimination methods. The PCA was applied to the standardized (MC and SNV) spectral data of the contaminated samples (8 substances, 11 levels of humidity) and the results are shown in Figure 6. The PCA separates the spectral data of samples into weakly distinguished groups containing spectra obtained for paper with different substances applied. Within each group, the data are distributed according to the level of relative humidity during the measurements—Figure 6a. The second derivative was applied to the spectral data (likewise in the case of the pure papers) in order to eliminate the effect of the variable relative humidity and clustering of the sample data contaminated with different substances—Figure 6b. The reduced influence of the linearly changing component of the spectra which corresponds to the water content in samples is revealed by the PCA applied to the second derivative. It separates the spectra into five well defined groups. Spectra of samples with soiling, characterized in the NIR region by a narrow, clearly observable bands (Paraloid B72, Vinavil Blue NPC, Butapren, linseed oil) are separated into well defined groups. In case of samples contaminated with substances characterized by a wide bands, additionally overlapped with these of the water absorption (white sugar-saccharose, Definol, methylcellulose,

Figure 6. First two principal components calculated for the spectra of Whatman paper covered with eight different substances (a) and for the second-derivative of the spectra (b) measured at 11 levels of h changing from 0.3 to 0.8.

gelatine), the spectra are clustered in a larger group and their discrimination is difficult or even impossible. This leads to the conclusion, that in case of substances characterized by the wide absorption bands overlapping with water bands in the NIR region the influence of variable humidity can not be effectively eliminated. 3.2.2 Influence of the variable humidity on identification of substances by means of spectral library In order to eliminate the influence of the paper properties during the identification procedure based on the NIR spectral library, the spectra of pure paper samples were subtracted from these of the soiled paper by applying autosubtract procedure supplied by the Grams software (GRAMS). This procedure uses an iterative algorithm, which determines the subtraction factor by minimizing the derivative of the resulting residual spectrum. The identification tests of substances applied to the paper samples were performed by comparing the “paper-subtracted” spectra with those available in the prepared library by using the library search program of the Grams. The comparison of spectra is performed by using the vector correlation method. The list of the closest matches is identified by ranking the library spectra with a calculated

181

LACONA8.indb 181

11/10/2010 5:37:24 PM

The observed influence of variations of the relative humidity on the spectra can be substantially reduced by the usage of analysis of the second derivative of the spectral data. Changes of the water absorbance of paper soiled with different substances influence markedly their identification possibility. In case of identification based on the spectral library data, the analysis of dependence of the search quality index on the relative air humidity indicates, that the low relative air humidity during measurement of the spectra represent the best identification conditions. Moreover, it was found that changes of the water absorbance characteristic of papers with presence of different types of soiling/sizing/contaminations can be used as an additional indicator for identification of the type of substance applied to the paper. Further research aimed on extension of the analysis reported here is in progress and results will be reported soon. ACKNOWLEDGEMENTS Figure 7. Dependence of the matching quality index (Q) vs. relative air humidity for three selected substances.

matching quality index Q with values ranging from 0 to 1, and the spectra are considered to be identical for matching index of 0. The best match between the examined and the library spectra corresponds to the lowest value of Q. Figure 7 shows the values of the matching index calculated for paper-subtracted spectra of selected samples measured at different levels of h in the range from 0.3 to 0.8. An increase of the relative humidity is accompanied by an increase of the matching quality index value. It confirms the result presented on Figure 4 and indicating, that the soiling substances applied on paper modify its water absorbance. The relative change of the paper humidity at given relative air humidity (h) depends on the substance added to the paper sample. As a consequence, the spectra with subtracted paper component obtained under different RH conditions are disturbed by the variable water content which depends on the chemical composition of the additive. 4

CONCLUSIONS

Spectra measured in the NIR region for reference samples of the contemporary model papers and also for soiled samples prepared by immersion or surface application of different substances confirm, that the relative air humidity have a significant influence on the results of statistical analysis of the spectral data and on the sample discrimination possibility.

This research work is supported by the Ministry of Science and Higher Education under the project No 2059/B/T02/2007/33. REFERENCES A. Candolfi, D.L. Massarta and S. Heuerdingb. Investigation of sources of variance which contribute to NIR-spectroscopic measurement of pharmaceutical formulations, Analytica Chimica Acta 345, 185–196, 1997. GRAMS/AI 8.0—Thermo Scientific (www.thermo.com) H. Yonenobu, S. Tsuchikawa and K. Sato. Near-infrared spectroscopic analysis of aging degradation in antique washi paper using a deuterium exchange method, Vibrational Spectroscopy, Volume 51, Issue 1, 100–104, 2009. J. Workman, Jr. L. Weyer, Practical Guide to Interpretative Near-Infrared Spectroscopy, CRC Press, Taylor & Francis Group, Boca Raton, London, New York, 2007. LU Guo-quan, HUANG Hua-hong and ZHANG Dapeng Application of near-infrared spectroscopy to predict sweetpotato starch thermal properties and noodle quality, J Zhejiang Univ SCIENCE B, 7(6): 475–481, 2006. M. Ali, A.M. Emsley, H. Herman and R.J. Heywood, Spectroscopic studies of the ageing of cellulosic paper, Polymer,Volume 42, Issue 7, 2893–2900, 2001. M. Blanco, M. Alcaláa and M. Bautistaa. Pharmaceutical gel analysis by NIR spectroscopy: Determination of the active principle and low concentration of preservatives, European Journal of Pharmaceutical Sciences Volume 33, Issues 4–5, 409–414, 2008. T. Næs, T. Isaksson, T. Fearn and T. Davies. A User-Friendly Guide to Multivariate Calibration and Classification, Chichester, UK: NIR Publications, 2002.

182

LACONA8.indb 182

11/10/2010 5:37:24 PM

Monitoring, imaging and documentation of artwork

LACONA8.indb 183

11/10/2010 5:37:25 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Experimentation of a three-focal photogrammetric survey system as non invasive technique for analysis and monitoring of painting surfaces decay condition P. Salonia, A. Marcolongo & S. Scolastico CNR, Institute for technology Applied to Cultural Heritage, Rome Research Area, Rome, Italy

ABSTRACT: This is a contribution to the process of monitoring and evaluation of frescoes laser cleaning, quantitative pictorial material loss detection, and qualitative data acquisition, necessary to evaluate the state of conservation of pictured surfaces (decay typologies, physical characteristics, morphology, etc.), useful to manage conservation interventions. We wished to verify the efficiency of an innovative survey system, based on the acquisition and processing of multi-view high quality images sequences, as an innovative and non invasive inquiry instrument that can support traditional techniques (like visual and instrumental analysis) in collecting colorimetric and morphologic data of paintings needing restoration, and for superficial decay kinetics analysis. 1

INTRODUCTION

Laser techniques applied to frescoes cleaning in art conservation have demonstrated very promising applications for restoration purposes. Laser based methodologies have had successful results, taking advantage of the peculiar characteristics of laser radiation for elemental composition, structural defects detection, and professional restorers are being acquainted with these new instruments and methods (Appolonia, L. & Brunetto, A. 1999). So far laser is a selective tool and the shotted beam is absorbed by dirt without affecting the substratum, as the beam power can be varied according to the operator’s needs, nevertheless it is of main importance to project the restoration intervention and document the artifact before and after the restoration process. Moreover degradation patterns must be recognized, recorded, and compared with similar ones for further investigations. Monitoring of the conservation degree of art pieces is essential in Cultural Heritage management and, as it allows to detect possible degradation phenomena and their diffusion, it requires repeated measurements of relevant parameters. In the field of interest of architectural surfaces frescoes and paintings, two of the most important parameters concern geometric and colorimetric information. To enable the detection of surface changes at a given resolution (e.g., erosion, mould growth, chemical alterations), geometric and colorimetric measurements must be sufficiently accurate and strictly correlated each other. Technologies adopted must then give accurate data during frescoes analysis and must be a useful

monitoring tool, before, during and after restoring intervention. Taking into account this purpose a trifocal digital photogrammetric-based technology has been tested for the acquisition of Quart Castle’s frescoes in Aosta. Medieval Quart Castle near Aosta is formed by a set of buildings arranged within a fortified perimeter, adapting to the natural contour of a difficult rocky slope. Medieval frescoes surveyed decorate the internal walls of the donjon. Actually they are being restored with laser based cleaning techniques by Anna Brunetto (Brunetto, A. 2004), under direction of the Superintendence of Valle d’Aosta Region. It has been used a laser type Nd JAG, 1064 nm wave length at different pulse time (SFR, LQS, QS). The trifocal digital photogrammetric-based technology have been previously tested in Siena at St. Maria della Scala, Cappella del Manto (Fig. 1), for the survey of frescoes during the cleaning intervention (Brunetto, A. 2008); relevant outputs after data processing have encouraged to adopt and apply this system widely in paintings documentation and restoration intervention. By means of the used survey technology, that is a multi image matching system, it has been possible to metrically reconstruct, at different LOD, from macro to micro scale, the all frescoes and samples of pictorial surfaces, obtaining accurate 3D scans (point clouds including both spatial and RGB information) that show all those information needed for the complete paintings knowledge. Establishing protocols for the acquisition and the 3D models reconstruction process (before,

185

LACONA8.indb 185

11/10/2010 5:37:25 PM

Figure 1. 3D RGB point clouds general 951000 points and detail of 2 million points.

during and after restoration) reproducibility, within the same data collecting and processing is a preliminary condition. Results of color measurement of a test point on a painted surface may change significantly when moving the capture device position. This technology fulfill this task, for time monitoring the kinetics of individuated decay pathologies, or even for controlling the effects of executed restoration interventions. A correct approach to artworks conservation requires the identification of defects at an incipient stage. Through the use of algorithms for surfaces comparative analysis, it has been possible to evaluate the effects of paintings laser cleaning and to register the micro variations in paintings depth in terms of amount of removed material and loss of pictorial fragments. Final precision depending on the distance between camera and object, on the type of lenses and on general photo conditions (object lighting). 2

and the digital camera calibration parameters, that have to be sent to the software for data processing, are necessary in order to allow the spatial reconstruction of camera centre position and to know the distortions due to the optics employed. The trolley allows to move and to secure the camera in different fixed positions on the bar in order to acquire sequences of images of the same object from different angle-shot. To produce a single 3D model, a sequence of three images, has to be taken from the left to the right, shifting the camera along the bar. The left and the right shots must be symmetric compared to the central shot, and the distance between them (the baseline) has to be carefully evaluated in relation to the optimal distance of the camera from the object (1:8–1:10), survey accuracy and level of detail required. There is no need of topographical support points, in order to create the single 3D model. However it is possible to make use of ground control points, during image processing, in order to geo-reference the single point clouds, in relation to a global datum system, and then to facilitate point clouds registration necessary for producing a final complete 3D model of the surveyed object. The systems satisfies characteristics of great flexibility and ease of use and guarantees, at the same time, accuracy of the geometric data acquired; however, using an image processing algorithm for 3D reconstruction, the system has some limits of application in relation to the characteristics of

TECHNOLOGY AND METHODOLOGY

2.1

Technology: Hardware

Concerning the frescoes survey, the trifocal digital photogrammetry-based survey technology adopted is ZScan Survey System and ZScan Micro (manufactured by Menci Software of Arezzo): they can be used to obtain RGB point clouds and relative 3D models, at different levels of detail, starting from the treatment of a number of images, taken with a limited set of constraints, through the use of a special acquisition equipment, and processed in a specific software, through the application of image matching algorithms. The acquisition equipment consists of a calibrated aluminum bar, of 100 cm (Fig. 2) for medium scale acquisition and of 30 cm (Fig. 3) with millimetric camera step positions, computer controlled, for micro evaluation. The bar can be easily mounted on a photographic tripod, which is provided with a small trolley for supporting a digital calibrated camera. Both the bar

Figure 2.

ZScan calibrated bar and digital camera.

Figure 3.

ZScan micro with automated bar.

186

LACONA8.indb 186

11/10/2010 5:37:25 PM

measured object surfaces. It reveals some limits in the 3D point cloud reconstruction of surfaces endowed with homogeneous colors, repetitive patters or high reflective materials. 2.2

Methodology: Data acquisition

The frescoes at Quart Castle have been surveyed in two survey campaign (July, September 2008) and in both occasions in two steps: from a distance of 1–2 m with a 24 mm calibrated lens with the 1 m bar for full coverage of the walls (a 10.2 megapixel CCD digital camera) and within a distance of 24–28 cm from the painted wall with a 80 mm calibrated macro optic (a 15.0 megapixel CCD digital camera) for details. From a distance of 150 cm the captured field of view has been of about 70 × 100 cm2 with taken images of 3872 × 2592 pixels (Fig. 4); from a distance of 24 cm have been framed 3,8 × 5 cm2 areas of 3888 × 2592 pixels (Fig. 5) using the micrometric bar, for close survey of the frescoes surface. The aim was to document wall paintings before, during and after the surface descialbo intervention and to evaluate the quantity of “material” removed by comparing elevation values (DEM), of correlated points before and after the restoration intervention in order to numerically and graphically describe surface’s trend, or to compare surface’s differences in time. This documentation has been then mainly a test to verify ZScan technology as a valid tool to control frescoes laser cleaning, but even if a detailed decay conditions analysis and pathologies record have not been done, nevertheless results obtained encourage to use this technology also for these purposes. It is possible in fact to query the detailed point cloud with ZMap tools so to extract dimensional data and eventually to upload any post processed output (vector or raster) in dedicated GIS (Salonia, P. et al., 2005). Frescoes status of conservation could then be periodically verified and recorded, along with the

Figure 4.

Figure 5.

70 × 100 cm frame.

3,8 × 5 cm frame.

probable causes of possible future decays, taking into account the complexity of decay phenomenology on pictorial surfaces. Wall frescoes are especially sensitive to temperature and humidity variations. Temperature and humidity act as shearing forces between layers or portions of a wall, weakening the material during the numerous cycles of these forces and causing the appearance of micro-cracks, debonds or anomalous strains on the surface, finally resulting in the decay of the artworks, actually Quart’s frescoes are interested by cracks and fractures occurring on walls, contour scaling, mechanical damaging, salt crusts. A painting on a wall can be considered as a layered structure with a support, coated with plasters, which serve as a base for the painting. These layers are less thick and more fragile than the support. Expansion and contraction of the support due to daily fluctuations of ambient parameters can produce large strains and eventually cracks in the layers, as they become less flexible with age. Furthermore, abrupt changes of temperature and humidity, and heat exposure may also cause unpredictable stress distributions in the heterogeneous materials of the support with consequent damage of the painted surface. All these mechanisms may lead to the formation of detachments that must be geometrically monitored. Some GCPs have then been taken with a reflectorless total station, registered in a local coordinate system, so to fix constraints for the 3D final model reconstruction (Fig. 6). By means also of topographic survey it has been possible to understand how the presence of the support cracks or discontinuities alter the movements of the painted surfaces. For what concern colorimetric measurements and comparison is of main importance to reproduce the same measurement conditions in time, test point localization, color temperature of lights (LUPO quadri-light 4 × 55 W 5400°K) and surveying geometry, in terms of spatial position between instrument and scene, have then been recorded.

187

LACONA8.indb 187

11/10/2010 5:37:26 PM

Figure 6.

Topographic survey and merged 3D model.

Figure 7.

LabCH values.

rectification and feature matching, in order to eliminate geometrical and optical distortion; selection of the image Areas Of Interest (AOI) that have to be processed; definition of the step resolution value, measured in pixel unit; production of a point cloud relevant to each group of three images, through the application of matching algorithms during image processing. Moreover, contextually to the point cloud production, the software allows to automatically create a texturized triangulated surface, through a triangulation process of the point cloud. According to the survey precision and to the representation detail needed, a step resolution value of 3 pixel was adopted for the elaboration of each three images sequences of global and detailed paintings, corresponding respectively to 0.654 mm point to point sample spacing and 0.037 mm on the created point cloud. Each sample of the global fresco (70 × 100 cm2) is around 0.9 millions of points and each micro (3,8 × 5 cm2) is about 0.74. These high resolutions permit to easily detect and measure features with advantages of

Each point of point clouds has a vertex color value that have to be eventually equalized to real colorimetric coordinates The Chemical Analysis Laboratory of the Superintendence of Aosta has taken spot colorimetric measurements (LabCH values) (Fig. 7), on some of surveyed areas, with a Minolta CR700 (technician Dario Vaudan). Those data can be used to correct images color in the 3D point clouds reconstruction. 2.3 Methodology: Data processing

Figure 8.

Triplets processing.

Approach in the surfaces reconstruction has been a two step approach; first the modeling of architectural structures and global frescoes, second the processing of micro details with high resolution, geo-referenced to the global one. Data processing has been carried out through the use of two dedicated software that are part of ZScan survey system. Single point clouds models have been extracted from each triplet of acquired images (Fig. 8). after chromatic equalization between different shots through the use of an image processing commercial software (Photoshop). The procedure for RGB point clouds extraction, consists of four main steps: images rectification, through the application of trinocular

Figure 9.

RGB point cloud.

188

LACONA8.indb 188

11/10/2010 5:37:27 PM

Figure 9a.

Figure 10.

Fissures measuring.

Figure 11.

Point clouds before and after descialbo.

Figure 12.

GCPs correlation.

Erosion measuring.

photo-realistic visualization. For instance it has been possible to quantify fissures dimensions, superficial alterations and features induced by material loss with millimetric accuracy; fissures vary from 1 up to 7 mm of depth and 150–180 cm of length (Figs. 9–9a). Other geometric measurements, angles, radius, cross sections are easily extracted from data using dedicated tools as those in RapidformTm (Inus Technology) software (Fig. 10). The study of deterioration processes implies availability of models simulating the deterioration dynamics, DEM analysis and comparison in ZMap permits to numerically and graphically describe surface’s trend, or to compare surface’s differences in the time (Fig. 11). The detection process aims at deriving both coverage and morphological measures regarding decay areas. Thus, it addresses both accurate location estimation and efficient shape segmentation. Single areas are overlapped trough omologous non collinear points (Fig. 12). Although in a noiseless situation three points is the absolute minimum, more points must be used in practice to ensure

that the match is more robust (i.e., less sensitive to uncertainty in the positions of the points). The match is then defined in the least-square sense and at least 6 GCPs have been processed with average residual values in xyz of 0.046 mm (Fig. 13). Once matched two or more pictures, taken with Zscan micro, in different times, DEM are generated and uploaded inside the comparing interface of the software. Z min, Z max, and range of values variation are automatically generated and reported (Fig. 13). We have evaluated after scialbo laser cleaning an average Δz of 0.078 mm, value that represent the average scialbo thickness. It has been then validated the potential and the efficiency of the cleaning method and technology adopted. Moreover, as throughout the cleaning process, some parameters such as the laser pulses are modified resulting in the removal of patina layers

189

LACONA8.indb 189

11/10/2010 5:37:28 PM

In future survey campaign of Quart’s frescoes will be possible to go deep in ZScan application, being this a reliable non invasive technology for a wide range of different data typology acquisition. ACKNOWLEDGEMENTS

Figure 13.

DEM analysis.

differing in thickness, each cleaned strip should be recorded between adjacent spots in Δt. It has been also possible to evaluate the percentage of area covered by patina layer, 60%, and their average size and spatial distribution on the overall fresco. 3

CONCLUSIONS

In this article have been presented the application of digital imaging techniques and technologies in the monitoring of laser cleaning intervention on wall frescoes. The aim was to evaluate the exact quantity of removed material trough the reconstruction and comparison of 3D models of wall surfaces before, during and after restorers work. Data extracted from 3D RGB point clouds have been extremely useful to validate laser cleaning methodology for frescoes restoration as submillimetric measures on 3D models can be taken due to high LOD obtained with ZScan Micro technology. Image Processing (IP) techniques can then be used for extracting information regarding the demaged areas in paintings, moreover, the proposed non-destructive approach enables measurements on the intensity distribution of decay, which is directly associated with the thickness of the patina crusts at these areas. The comparison of results derived by the monitoring of cleaning processes in time provides an overall assessment of the potential and the limitations of optical digital 3D inspections also in the reliable estimation of surfaces decay kinetics.

It would not have been possible to accomplish this presented work without the remarkably intelligent and wonderful availability constantly shown by restorer Anna Brunetto. With outstanding professional far-sightedness she accepted to submit to our micro-metric survey her interventions of laser cleaning for both Cappella del Manto at St. Maria della Scala Church—Siena, and Quart Castle—Aosta. Authors wish to thank Lorenzo Appolonia, Chief of Research and Co-Funded Projects Head Office at the Superintendence of Cultural Heritage of Aosta, autonomous region of Valle d’Aosta, for the usual availability and collaboration in every activity. A special thank is also due to his contributor Dario Vaudan, from the Laboratory of Chemical Analysis of the Superintendence, for the colorimetric data that he had collected and elaborated. Last but not least, authors wish to thank their friends Francesca Ceccaroni and Luca Menci, from Menci Software Ltd, Arezzo, along with their valid colleagues for their precious availability and permission to the usage of Zscan Micro equipment. REFERENCES Appolonia, L. & Brunetto, A. 1999. La scelta dell’utilizzo del sistema LASER per la pulitura del Priorato di Sant’Orso in Aosta. In: Materiali e tecniche per il restauro. Cassino 1–2. Brunetto, A. 2004. L’utilizzo della strumentazione laser per la pulitura delle superfici nei manufatti artistici. Collana i Talenti. Saonara (PD): Il Prato. Brunetto, A. 2008. Il laser per la rimozione delle scialbature dalle pitture murali della Cappella del Manto in Santa Maria della Scala a Siena. In Il colore negato e il colore ritrovato: 193–204. Firenze: Nardini. Salonia, P. et al. 2005. L’esempio del ciclo di affreschi alto-medioevali della Collegiata di Sant’Orso in Aosta: Tecnologie GIS a supporto del progetto di conservazione. In: XXI Convegno Internazionale Scienze e Beni Culturali. Bressanone.

190

LACONA8.indb 190

11/10/2010 5:37:30 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

RGB-ITR: An amplitude-modulated 3D colour laser scanner for cultural heritage applications R. Ricci, L. De Dominicis, M.F. De Collibus, G. Fornetti, M. Guarneri & M. Nuvoli ENEA, Frascati, Rome, Italy

M. Francucci ENEA fellow, Frascati, Rome, Italy

ABSTRACT: The RGB Imaging Topological Radar (RGB-ITR) is an amplitude-modulated, 3D colour laser scanner entirely designed and developed at ENEA laboratories. The system incorporates three laser beams (450 nm, 532 nm, 650 nm) in monostatic configuration, and can simultaneously record range and self-registered RGB colour data. Such distinguishing feature makes the RGB-ITR sensibly different from currently available 3D scanners, where colour imaging information is acquired independently and must be mapped via software onto the naked 3D model. The RGB-ITR specifically addresses the requirements of cultural heritage applications, especially in terms of non-invasiveness, accuracy and versatility. The system has already been used with success for the 3D digitisation of the frescoed medieval S. Trinity church in Hrastovlje (Slovenia) and of the Carafa Chapel in Rome (Italy), a masterpiece of Italian Renaissance. The results of these campaigns are reported. 1

INTRODUCTION

Laser range finding (Blais 2004) is nowadays a mature technology, widely applied in the cultural heritage domain for the 3D digitisation of rare, endangered or hardly accessible pieces of art (Blais & Beraldin 2006, Ricci 2008a). Laser 3D scanners can significantly improve computer-aided conservation and restoration practices, especially when used in combination with other imaging diagnostic tools (Ferri De Collibus et al. 2005, Ricci et al. 2008b). The purely geometric output produced by the vast majority of currently available laser scanners, though, is not enough for most cultural heritage applications. It is common practice to superimpose, onto the naked 3D models generated by most devices, textures derived from ordinary digital photos. Since the generation and superimposition of texture images is realised via software, this technique has many drawbacks and limitations in terms of attainable visual quality and accuracy. In order to meet the demanding requirements posed by restorers and conservators, the ENEA Artificial Vision laboratory in Frascati (Rome) designed and developed the RGB Imaging Topological Radar (RGB-ITR), a new 3D scanner with both accurate range finding and intrinsic colour imaging capabilities. The scanner uses three Amplitude-Modulated (AM) laser sources—in the red, green and blue region of the visible spectrum

respectively—and can thus acquire in a single scan both range and self-registered colour information. Even though still a laboratory prototype, the RGB-ITR has already been tested in the field with very good results. In this article we provide an introductory description of the system and report the results of measurement campaigns carried out on two important masterpieces of European artwork, namely the S. Trinity church in Hrastovlje (Slovenia) and the Carafa Chapel in Rome (Italy). The article ends with an anticipation of future planned developments and conclusions. 2

PREVIOUS WORK

3D colour laser scanning is still, at the moment of writing, mostly confined in research laboratories. To authors’ knowledge, the only 3D scanner with native colour acquisition capabilities is the autosynchronised High Resolution Colour Laser Scanner (HRCLS) developed by the Visual Information Technology Group of the National Research Council of Canada (NRC-CNRC). A first prototype of the HRCLS, which uses optical triangulation for the determination of range, was realised by the NRC-CNRC for the digitization of paintings (Taylor et al. 2003). The system was used to evaluate the performance

191

LACONA8.indb 191

11/10/2010 5:37:30 PM

3.1

ITR’s characteristics and functioning principle

requirements of a 3D acquisition system for the recording and documentation of paintings (Blais et al. 2005). Three (red, green, blue) laser sources were combined into a single-mode optical fibre, projected onto the target and imaged by a linear CCD after splitting the three colour components by means of a prism. The scanning head was mounted on a translation stage that enabled to scan small portions of 4 cm × 20 cm of the painting. In the configuration adopted, the system provided a lateral spatial resolution around 50 μm and a depth uncertainty of 10 μm on flat surfaces. A similar system was employed in October 2004 to realise an accurate digital reconstruction of the famous Leonardo’s “Monna Lisa” (Blais et al. 2007). A total of 72 sequential bands for the obverse side and 68 for the reverse side of the painting were recorded, stitched and merged by software to form the complete colour model. Arius3D Inc., a small service company headquartered in Mississauga, Ontario Canada (http://www.arius3d.com), obtained an exclusive license for this technology from NRC-CNRS and is now realising on demand colour laser scanners based on the same functioning principle. The position along the Z-axis (the laser axis) is measured by laser triangulation, enhanced by the application of synchronized scanning geometry. This method uses one side of a galvanometer-driven scanning mirror to deflect the laser across the scanned object, while the opposite side of the same mirror is used to cancel the return beam’s angular movement across the CCD sensor. A sub-pixel interpolation scheme is used to enhance the CCD. The HRCLS, though extremely accurate in optimal operation conditions, is affected by the same limitations of monochromatic 3D scanners that use optical triangulation for the determination of range. In particular, the system can only be used for the 3D digitisation of relatively small objects— the maximum scale being determined by the baseline distance of the laser source from the CCD.

where fm is the modulation frequency and v the velocity of light in the transmitting medium. For laser optical powers such that the signal shot-noise dominates over all other noise sources in the detection process (typically, a few nW at the output of the detection fibre), the accuracy of measurements can be showed to increase with the modulation frequency fm, (Nitzan et al. 1977):

3

σR ∝

THE RGB IMAGING TOPOLOGICAL RADAR

The RGB-ITR is the last offspring of a series of amplitude-modulated, monochromatic 3D scanners realised at the ENEA Artificial Vision laboratory (Frascati, Rome), and collectively identified by the acronym ITR (Imaging Topological Radar). In the following subsections we briefly introduce the functioning principle and design characteristics of all ITR systems, followed by a high-level architectural description of the RGB-ITR.

All ITR systems are based on the Amplitude Modulation (AM) range finding technique and share a common modular design, comprising an active and a passive module. The passive module coincides with the system’s optical head, and basically includes the transmitting and receiving optics. The active module is composed of the laser source(s) and detector(s). The two modules are physically separated and optically connected by means of optimised optical fibres. This enables one to use the system in hardly accessible or even hostile environments (Bartolini et al. 2000) without compromising its performances. Efficient noise rejection is obtained by using narrow field-of-view, interferential filtering and low-noise detection electronics. Self-occlusions, as well as off-axis aberrations, are inherently disposed of, due to the monostatic configuration of the launch and receiving optics. In monochromatic ITR systems, a single low-power (∼10 mW) diode laser source is used, whose optical wavelength typically lies in the near infrared part of the electromagnetic spectrum (∼800 nm). In the AM rangefinding technique the laser beam is exploited as the carrier of a radiofrequency signal, which modulates the beam intensity. Distance is determined indirectly from the phase delay Δϕ of the collected signal photocurrent with respect to the reference signal used to modulate the source, according to the formula: d=

vΔφ , 4π fm

1 m fm SNR N i

(1)

(2)

In (2), m is the modulation depth and SNRi stands for the current signal-to-noise ratio, SNR NRi =

Pητ hf Γ

(3)

which depends on the laser optical frequency f, the integration time τ, the detector’s quantum efficiency η, the overall optics merit factor Γ and the collected power P.

192

LACONA8.indb 192

11/10/2010 5:37:30 PM

Because of phase periodicity, the AM range finding technique is affected by the so-called “folding” ambiguity. The system returns one and the same distance value for target points, whose relative separation along the line of sight is a multiple of half the modulation wavelength. In order to overcome this problem, the laser probe is modulated at two different frequencies, whose values are taken far apart from each other. The lower modulation frequency is determined by the requirement that the corresponding measurement range encompasses the whole scene of interest. Lowfrequency measurements are then used to remove the ambiguity that affects the corresponding highfrequency, more accurate measurements. By using this technique, it is possible to achieve submillimetric accuracy on targets laying at distances from a few tenths up to several tens of metres, without the need of scaffolding. 3.2 RGB-ITR: Shape and colour Profiting of the experience maturated over a decade with monochromatic ITRs, the Artificial Vision laboratory recently managed to realise the RGBITR, a colour 3D laser scanner, which enables the simultaneous recording of both range and colour, in the form of RGB triplets. To authors’ knowledge, this is the first AM 3D scanner ever realised that can natively acquire colour information. The system (Fig. 1a) uses three pigtailed laser sources at optical wavelengths of 450 nm (blue), 532 nm (green) and 650 nm (red) respectively, combined in one of two possible different optical setups. In the first setup, the three beams are connected to the optical head via single mode optical fibres, kept in a fixed position by means of a patented mechanical device (ferrule) that is placed in the launch section of the optical head (Fig. 1b). A single 40 mm-focus, diffraction-limited, achromatic lens (Melles-Griot LAL-011) focuses the slightly separated beams onto the target by means of a motorised, two-degrees-of-freedom rotating mirror, with the red beam directly impinging on the mirror centre, and the blue and green beams diverging by a total angle of 6.045 mrad on the same plane. The backscattered signals are focused by a 50.6 mm-diameter doublet lens (Newport P086) onto three multimode optical fibres (1 mm diameter) in the receiving ferrule. In the second setup, the three beams are firstly combined into a single ray by means of a dichroic optical component, connected to the optical head via a single-mode optical fibre. The same lens and mirror as in the previous setup are used to focus and sweep the beam onto the target, and the collected backscattered signal is then split again into

Figure 1. (a) RGB-ITR in operation and (b) scheme of the ferrule-based optical setup.

the three original red, green and blue components by means of an optical demultiplexer. In both cases, the red, green and blue return signals are separately detected by three low-noise avalanche photodiode detectors, and analysed by means of three Stanford SR-844 lock-in amplifier units, which are also used to modulate the laser sources. Although all the beams are modulated, only the red and blue backscattered signals are exploited for, respectively, high-frequency and low-frequency range determination. Typical modulation frequencies are 190 MHz for red, 1–10 MHz for blue and 25 kHz for green. The three colour channels are converted in standard RGB form by using the Grassmann law (Schanda 2007)—which states the linearity of chromatic response in human colour perception—with coefficients corresponding to the optical frequencies of the laser sources used. The RGB triplets are then subject to a calibration procedure that enables to fix the white point in colour space by means of data preliminarily acquired on a white certified diffusive target. More details on the colour calibration procedure are given in the next section.

193

LACONA8.indb 193

11/10/2010 5:37:31 PM

Specifically developed, copyrighted software applications are finally used to process range and calibrated RGB data and generate faithful, highly realistic 3D digital models. The resulting models can be further post-processed (purged from spurious vertices and outliers, registered in a single reference frame etc.) and exported into the most common 3D file formats. 4

COLOUR CALIBRATION

In this section, we examine the various factors that influence the colour measuring capabilities of the RGB-ITR, in order to assess the quality of the colour information that the system is ultimately able to produce. Since these aspects are still object of active research at the moment of writing, the analysis is carried out at a qualitative level. For any sampled surface point, raw colour data returned by the RGB-ITR simply consists of a triplet of voltage values. Each value represents the (amplitude of the oscillating) red, green or blue light power reflected by a particular surface pointlike area, as collected by the receiving optics and revealed by the detector. The reflection of light by a generic surface is described in terms of the surface Bidirectional Reflectance Distribution Function (BRDF), a generally complicated function that specifies the ratio of reflected radiance exiting along a certain direction Ωout to irradiance incident from direction Ωin (Nicodemus 1965). The BRDF strongly depends on the surface microscopic characteristics, but for a typical real—i.e. neither fully specular nor fully Lambertian—surface, it can be approximated by the weighted sum of three terms, related to specular, Lambertian and retroreflected radiance components respectively. Since the optical axes of the RGB-ITR’s launch and receiving stages coincide—i.e. the system is monostatic-, only radiation reflected in a small solid angle around the incidence direction is revealed. So—neglecting the incidental case of strictly normal incidence, where the specular component also plays a role—the RGB-ITR is only sensitive to the Lambertian and retroreflected components. Since the contribution of the retroreflected component is usually small— apart from a minor category of very special surfaces (retroreflectors)—we focus our qualitative analysis on the Lambertian part. Lambertian reflection is isotropic, that is, exiting radiance I is constant over the whole hemisphere. At large distances z, the power falling on the receiver is roughly equal to: _ P∼

I R S cos θ , z2

(4)

where R is the area of the receiver and S is the area illuminated by the laser spot on the target, which varies with the angle θ formed by the receiver axis with the normal to the target surface. Ideally, since S = S 0/cosθ, the dependence on θ should cancel out. In practice, owing to the optical efficiency of the receiving optics, the RGB-ITR could produce different outputs for the same surface point if the orientation of the surface with respect to the receiver axis changes, even if the distance of the surface from the receiver remains the same. Distance also obviously influences colour measurement. In first approximation, as shown in Equation 4, the received power roughly depends on the inverse of distance squared. A correction of the raw colour data taking into account this trend can in principle be made, owing to the fact that distance is also recorded by the system, but other, more subtle factors must also be considered. These factors are related to possible misalignments of the launching and receiving optical axes, and to the combined effect of the receiving optics efficiency with the varying dimension of the spot on target. Each of the three laser beams can be assumed to be Gaussian with good approximation, corresponding to the TEM00 component in an expansion of the beam in transverse modes (Pampaloni & Enderlein 2004). The spot of a Gaussian beam on a plane normal to the propagation direction has a minimum at a certain position z0 along the beam— beam waist, S0 = S(z0)-, and diverges from that position in either direction as 2

⎛ z z0 ⎞ S ( z ) = S0 1 + ⎜ , ⎝ zR ⎟⎠

(5)

where zR π w02 λ is the Rayleigh range, depending on the optical wavelength λ. By focusing the beam onto a certain point on the scene, the waist is moved at that point, that is, z0 is made to coincide with the distance of the on-focus target point from the nodal point of the launching stage. Portions of the scene, whose distance over the beam line is greater or less than z0, are out of focus. When falling onto those regions, the laser spot S(z) > S0 illuminates larger areas of the target surface. In particular, when z < z0, the image onto the receiver of the target area illuminated by the spot can be larger than the area of the receiver, which thus collects only part of the reflected radiation. An even greater source of detected power loss is the misalignment of the launching and receiving optical axes, which can have unpredictable, deleterious effects at all distances, depending on the mutual orientation of the axes.

194

LACONA8.indb 194

11/10/2010 5:37:32 PM

These effects are illustrated in Figure 2, where we report the dependence on distance of raw colour measurements performed on a certified white diffusive target (Spectralon® STR-99-020, diffuse reflectance 99%). The on-focus distance corresponds in this case to z0 = 5 m, where all the curves have a maximum. For z > z0 the collected power falls down, mainly because of the misalignment of the optical axes and the 1/z2 trend. For z < z0, this trend is counterbalanced by the fact that the receiver intercepts only part of the reflected light, so the collected power also falls down. As a consequence, all the curves have a bell-like shape. This clearly indicates that simply correcting the raw data to get rid of the 1/z2 dependence might not always produce acceptable results. Another factor to be considered is the instability of the laser sources over time. This is illustrated in Figure 3, where we report, for each channel, the normalised power returned by a retroreflector mounted on the back of the scanning mirror, and measured at fixed intervals of 40 minutes during the scan. It is immediately apparent that these values are not constant over time. The temporal drift of the laser power, more evident for the green source, introduces a systematic error in the determination of colour, since at different times the system is likely to measure different colours of the same target point in otherwise identical conditions. All these factors are to be taken into account for a correct calibration of colour information. Ideally, the colour content of a 3D model representing a real target should reflect the intrinsic properties of the target surface. This statement is equivalent to the requiring that two colorimetrically indistinguishable regions of the surface are represented in the model as having the same colour, independently on the conditions under which those regions

Figure 3. Normalised colour detected signals returned by a retroreflector mounted on the back of the scanning mirror, and measured at fixed time intervals of 40 minutes.

were sounded during the scan. This requirement, far to be satisfied by 3D scanners that make use of ordinary CCDs for the recording of colour, can in principle be met by the RGB-ITR, provided an accurate calibration is carried out. At the moment of writing, the quantitative aspects of the colour calibration procedure are still under study, but the qualitative analysis exposed in this section suggests that such an optimal calibration is indeed possible for the RGB-ITR. This opens unprecedented application scenarios, especially in the cultural heritage domain, where the colorimetric characterisation of a figurative artwork is as—if not more—important than the accurate determination of the geometric shape of the artwork itself. 5

RESULTS

The RGB-ITR has been already utilised in two 3D digitisation campaigns. The first campaign was carried out in Hrastovlje (Slovenia, 2007) under request of the Institute for the Protection of Cultural Heritage of Slovenia. The second campaign was performed in Rome (Italy, 2008) in collaboration with the Foundation for Worship Buildings, a Department of the Italian Ministry for Internal Affairs. By operating the system in the field, we were able to test its performances in realistic scenarios, in view of qualifying the RGB-ITR as a valid advanced tool for cultural heritage cataloguing and conservation.

Figure 2. Raw colour detected signals on a white certified diffusive target (Spectralon® STR-99-020, diffuse reflectance 99%) as a function of distance.

5.1

St. Trinity Church (Hrastovlje, Slovenia)

In the framework of the European Union Culture 2000 Community Programme, an Italo-Slovenian

195

LACONA8.indb 195

11/10/2010 5:37:37 PM

Figure 4. 3D colour model of the Holy Trinity church in Hrastovlije (Slovenia).

collaboration was set up with the objective to realise an accurate, colour 3D model of the interiors of the Romanic Holy Trinity church in Hrastovlje (Slovenia), for monitoring, restoration and cataloguing purposes. All the walls and vaulting of the church are decorated by frescos dating back to the XV century and representing scenes from the Old and New Testament. The paintings, remained hidden under plaster for a long time and rediscovered in 1949, are considered a masterpiece of Slovenian medieval mural artwork. During a five-day campaign in September 2007, 60% of the church’s surface was 3D-digitised by means of the RGB-ITR, for a total of 11 scans and almost 90 data recording hours. Figure 4 shows a snapshot of the 3D model, consisting of 9,720,000 vertices and 18,123,126 triangles. 5.2

Carafa chapel (Rome, Italy)

In early 2008, ENEA was requested by the organisers of “Il ‘400 a Roma”—a temporary exposition focused on Roman Renaissance masterpieces—to realise a complete 3D model of the Carafa Chapel in Rome. The chapel—located in the south transept of Santa Maria sopra Minerva, the only Gothic church still present in Rome—contains Filippino Lippi’s magnificent fresco of The Assumption and several other beautiful frescoes executed between 1488 and 1492. The intention of the organisers was to provide exposition visitors with a fully satisfactory multimedia enjoyment of this superb artwork, otherwise very difficult to appreciate in-situ because of the scarce illumination and the considerable distance of the frescoed walls and vaulting from the visitors’ viewpoint, constrained by the presence of a fence at the chapel entrance.

Figure 5. Complete 3D colour model of the Carafa Chapel (Rome, Italy).

In February 2008, a series of 6 scans, for a total of about 120 data recording hours, were performed in the chapel. From the recorded data, 6 very detailed, colour 3D meshes were produced, for a total of 40,250,232 vertices and 92,772,253 triangles. The partial meshes were subsequently registered and integrated in a single 3D model. In order to enable a smooth interactive visualisation, the model’s geometric complexity was greatly reduced by removing up to 90% of the vertices with an adaptive decimation procedure, while at the same time integrally preserving the original resolution of visual information. This was made possible by the 1-to-1 correspondence of range and colour data, which provides natural texture mapping coordinates and enables error-free texturing of imaging data even on decimated models. The 3D model of the Carafa Chapel (Fig. 5) was exposed in “Museo del Corso”, a private museum in the centre of Rome, for a total of 4 months, and produced considerable reactions in national media (newspaper articles, TV reports etc.). 6

CONCLUSIONS AND FUTURE DEVELOPMENTS

The RGB-ITR is the first AM 3D laser scanner that enables the simultaneous recording of submillimetric range and self-registered colour (RGB) information. This feature, combined with

196

LACONA8.indb 196

11/10/2010 5:37:37 PM

other characteristics such as non-invasiveness, versatility, transportability and possibility to operate at long distances without compromising performances, make the new system unique in the—quite crowded—panorama of 3D digitisers. We believe the RGB-ITR has the potential to significantly contribute to a wider adoption of advanced laser scanning technology in the cultural heritage domain, as also confirmed by the enthusiastic acceptance of artwork specialists during the Hrastovlje and Rome campaigns. Research is on-going to develop a more robust hardware and colour calibration methodology, so as to permit colorimetric measures aimed at the study of colour degradations over time and at pigment characterization. Future plans also include investigating the possible use of polarising filters to isolate the Lambertian component, normally used in colorimetric studies, from the rest of the backscattered signal.

REFERENCES Bartolini, L. Bordone, A. Coletti, A. Ferri De Collibus, M. Fornetti, G. Lupini, S. Neri, C. Riva, M. Semeraro, L. & Talarico C. 2000. Laser In Vessel Viewing System for Nuclear Fusion Reactors. In: International Symposium on Optical Science and Technology, Proc. SPIE 4124: 201–211. Blais, F. 2004. Review of 20 years of range sensor development. Journal of Electronic Imaging 13 1: 231–240. Blais, F. Taylor, J. Cournoyer, L. Picard, M. Borgeat, L. Dicaire, L.G. Rioux, M. Beraldin, J.A. Godin, G. Lahnanier, C. & Aitken, G. 2005. Ultra-High Resolution Imaging at 50 μm using a Portable XYZ-RGB Color Laser Scanner. In: International Workshop on Recording, Modeling and Visualization of Cultural Heritage, Centro Stefano Franscini, Monte Verita. Ascona.

Blais, F. & Beraldin J.A. 2006. Recent developments in 3D multi-modal laser imaging applied to cultural heritage. Machine Vision and Applications 17: 395–409. Blais, F. Taylor, J. Cournoyer, L. Picard, M. Borgeat, L. Godin, G. Beraldin, J.A. Rioux, M. & Lahanier, C. 2007. Ultra High-Resolution 3D Laser Color Imaging of Paintings: the Mona Lisa by Leonardo da Vinci. In: Castillejo et al. (eds), Lasers in the Conservation of Artworks; Proc. int. conf. LACONA VII, Madrid, 17–21 September 2007: 435–440. London: Taylor & Francis Group. Ferri De Collibus, M. Fornetti, G. Guarneri, M. Paglia, E. Poggi, C. & Ricci, R. 2005. ITR: an AM laser range finding system for 3D imaging and multi-sensor data integration. In: Proc. of ICST 2005 (International Conference on Sensing Technology, Palmerston North, New Zealand, 21–23 November 2005: 641–646. Nicodemus, F. 1965. Directional reflectance and emissivity of an opaque surface. Applied Optics 4 7: 767–775. Nitzan, D. Brain, A.E. & Duda, R.O. 1977. The Measurement and Use of Registered Reflectance and Range Data in Scene Analysis. Proc. IEEE 65: 206. Pampaloni, F. & Enderlein, J. 2004. Gaussian, Hermite-Gaussian, and Laguerre-Gaussian beams: A primer. ArXiv:physics/0410021. Ricci, R. 2008a. Three-Dimensional Scan of Underground Cavities. In: Fiorani et al. (eds), Laser Applications in Environmental Monitoring: 227–249. New York: Nova Science Publishers, Inc. Ricci, R. Ferri De Collibus, M. Fornetti, G. Francucci, M. Guarneri, M. & Paglia, E. 2008b. ITR: A laser rangefinder for cultural heritage conservation applications with multi-sensor data integration capabilities. In: Castillejo et al. (eds), Lasers in the Conservation of Artworks; Proc. int. conf. LACONA VII, Madrid, 17–21 September 2007: 447–452. London: Taylor & Francis Group. Schanda, J. (ed.) 2007. Colorimetry, understanding the CIE system. Wiley. Taylor, J. Beraldin, J.A. Godin, G. Cournoyer, L. Baribeau, R. Blais, F. Rioux, M. & Domey, J. 2003. NRC 3D Technology for Museum and Heritage Applications. The Journal of Visualization and Computer Animation 14: 3.

197

LACONA8.indb 197

11/10/2010 5:37:37 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

3D laser reconstructions of Buddhist temple from Ladakh D. Ene & R. Rădvan National Institute of R&D for Optoelectronics INOE 2000, Măgurele, Romania

ABSTRACT: Delicate features needs delicate approaches, multidisciplinary policy being the answer, offering more complex and complementary information. Mobile optoelectronical techniques where developed in the past years, techniques that allows in situ measurement with satisfactory results. This paper presents 3D laser scanning results of Buddhist temple from the northern part of India, Jammu & Kashmir. Location were the measurements were made was chosen due to the particularity of each location, a possibility of collapsing, restoration intervention that are in progress or will be made. It’s an example of laser used in conservation of artifacts in order to create digital optimum 3D models, results accessible in a web sharable format. 1

As a negative aspect of this mobile technique the digital models may be displayed only with laser wavelength reflected intensity, value that depends by many factors, including the distance between the scanner and the object and the object’s shape. For the 3D colored model photo images were overlapped on the 3D model. Common details on the photo image were identified with the cloud of points displayed also with the 690 nm reflectivity and after that regular processing workflow were followed. [Heritage & Large]. Its scanning location has its own particularity but the common point of all was the necessity to make an accurate digital models.

INTRODUCTION

In northern part of India, in the state of Jammu & Kashmir, is a region strongly influenced by Tibetan culture—Ladakh, also known as Little Tibet, due to geographical position and commercial routes that passes through Ladakh for centuries. A remote area, location were measurement were made are at altitude above 3 000 m, regions still facing with the lack of electricity. This scanning campaign comes at the initiative of Tibet Heritage Fund (THF) who needed high accuracy 3D digital models for the restoration/ conservation activities that the NGO coordinate at temples from Ladakh. Depending of the casuistic of the location and the limited time for scanning, 2 temples were chosen: Red Jampa Maitreya from the state capital Leh, roof’s temple that will be repaired and a supporting corridor, made in 20th century will be demolished. The second location is a temple complex from Alchi—Tsatsapuri with elaborated restoration intervention. [Alexander & Catanese, 2007]. Additional a third location was chosen, two cave temples, from Sasspol village, mainly due the risk of collapsing hill, where this caves are. 2

2.1

IN SITU ACTIVITIES

First item to consider in this campaign was the portability of the scanning equipment. Because of the diversity of the application, regions with high resolution (< 500 μm) combined with 3D model of buildings with medium resolution (∼ cm) a phase shifting single line (690 nm) 3D laser scanning were preferred. [Guarner, M. et al. 2008].

Red Jampa Maitreya

Red Chamba Lhakhang from Leh is a temple build between 1400–1440 years, when king was Tragspa Bumde. In 1840’s was seriously damaged, during Dogra invasions but soon restored. In the mid of 19th century, due to the water damage of the north wall was build a corridor around a central Maitreya statue leaving intact north and west wall. [Nicolaescu & Alexander] Scanning activities at this temple was focused on two objectives, high resolution scanning of the original mural paintings (north and west wall) mapped on the rest of the temple and to digital reconstruct the large Buddha Maitreya statue. To record in high resolution these two walls, dedicated recordings were made. For the rest of the temple a general scan was enough, allowing in this was to identify temple and Maitreya room entrances. The main problem was to digitize the north wall, because of the narrowed corridor, with an opening

199

LACONA8.indb 199

11/10/2010 5:37:38 PM

less than 1 meter, giving high values of the wall’s reflectivity in the lower part of the walls in contradictory with the upper part of the wall. For the digital reconstruction of the Maitreya statue, 6 different recordings, 3 with the scanning equipment placed at the bottom of the statue, and other 3 scans with the equipment placed at the same level with the statue’s head, benefit from pagoda’s window. 2.2

Alchi Tsatsapuri

At a distance of 70 km from Ladakh Leh is Alchi, village know mostly because of the Choskhor temples. Tibet heritage Fund coordinates restoration activities at another temple complex, Tsatsapuri.

Figure 1.

Red Jampa’s corridor scanning.

Figure 2.

Dukhang’s 2nd level pillars scanning.

This complex was build between 13th–15th century, consisting in 3 temples: Lama’s reception room, Stupa Hall and Ridzong Dukhang Chenrezig Lhakhang, temple displayed on two levels and a lantern, recent painted. Special attention was given to shape recording of the Dukhang’s 2nd temple pillars, specific of Buddhism architecture, which required high resolution scanning. As a negative aspect of the resolution increasing was the longer acquisition time needed. Besides scanning of the temples, dedicated recordings were made also to the central court of the complex, allowing in this way relative positioning of the temples. 2.3

Sasspol caves

Near Alchi, less than 10 km, is a small village Sasspol. In a collapsing hill and with the nearest road at 2 km, probably the hardest part of the one day scan was caring out the equipment (3D laser scanning, laptop, power generator with combustible, power adaptors and stabilizator) in the caves. In the good tradition of Buddhism cave temples, from which the ones from Sasspol can be enumerate, dated around 15th-16th century (based on iconography) and with a few information known about the history of this caves. [Toubeki & al 2009]. Laser scanning was made in two of the caves, the main temple, with good mural paintings conservation and a second temple, with a collapsed residential room and paintings partial retouched. For the main room 3 different scans where needed, because of the shape of the cave, a horizontal section being a pentagon with one of the corners a little elongated. Belong a general scan of the second temple, another recording was made with the entrance in the temple and with the residential room.

Figure 3.

Sasspol 1st cave-temple scanning.

200

LACONA8.indb 200

11/10/2010 5:37:38 PM

3

PROCESSING

Computing of the Ladakh scanning campaign data, raw information more than 60 GB, were made during several months, and still are not yet fully processed, due to the high information that are available. Especially at Alchi—Tsatsapuri, were due to the scans number, caused by the specific Buddhist elements like drums, pillars, scarves that make it almost impossible to record entire temple from one general scan (as Sasspol caves shape allowed). This high number of scans (Stupa’s Hall was reconstructed by 6 scans, 2nd level of Dukhang by 4 scans, additional 8 scans for the pillars) allowed high resolution scans, giving 3D models with rich details. Furthermore, the scenes were digitized preserving the objects in specific position. Architectural details required special attention adding more time in the processing, besides the normal workflow, from raw data to 3D model, computational time. Processing at the Sasspol caves faced with the high number of holes, caused by the rock’s shape shadowing. In order to obtain a complete digital model, but still preserving high level of details,

the caves were Poisson reconstructed, after 10% decimation. in the next stage the holes filled model was than volume merged with the original model, maintaining in this way the original data and filling also the gaps. [Kazhdan1 & al]. Same problem was to reconstruct the tsa tsa’s, more than 50 elements. Same workflow was applied in this case, obtaining 3D models with no imperfection, available for further application (inspection or 3D printing). Main aspect in the 1st temple was to obtain a 3D colored model, as much possible as it gets by photo mapping. The available 3D colored model is consist by entire west and south wall and specific scenes on the north wall (in order to keep areas with high resolution, we prefer to obtain colored model only on some scenes, instead of 3D colored whole wall with low resolution and rather flat surface). Same problem with the scarves was in Maitreya statue scanning, combined with the height and the small distance in front available justified the number

Figure 6. Horizontal section of Sasspol cave and the available 3D colored digital model. Table 1. Τsa tsa Poisson reconstruction parameters. Figure 4. Tsatsapuri complex- up view of the 3D digital models.

Figure 5. Red Jampa’s general view and the corresponding murals.

Parameter

Value

Octree depth Solver divide Samples per node Surface offseting

11 11 3 3

Figure 7. Tsa tsa reconstruction—original data and reconstructed data (zoom with a reconstructed tsa tsa).

201

LACONA8.indb 201

11/10/2010 5:37:39 PM

ACKNOWLEDGMENTS The authors would like to thank for the collaboration to Patrick Juergens and André Alexander, from Tibet Heritage Fund, and to Romanian restorer Anca Nicolaescu. Activities were developed during project IMAGIST 91009 PN II. REFERENCES

Figure 8.

Red Maitreya statue.

of the scans. This high number of recordings needed more time to filter and align the data in order to produce an accurate digital 3D model. 4

DISCUSSION

Main goal of the Ladakh 3D scanning was to produce accurate digital models of the Buddhist temples and to make these models accessible. For this several option were considered, including 3D colored models. So far, the most compact option was to integrate the reflectance in 3D models. Time for actual scanning was also considered, this time efficiency allowed 1 day measurements for complete shape recording of the caves from Sasspol. The UV reflectance map were generated in 3D.mov files and posted at the THF website, http:// www.tibetheritagefund.org/ under the section of Research—3D scans. The website will be continuous updated while new internet engines with 3D colored models will be available, elements that will allow simple 3D commands, like rotate, walkthrough, pan or zoom in or out. Results should not be seen only as a “cultural entertainment” or ordinary documentation tool but more as a study on stability/dynamics of the building will be included. This model may be considered as a common reference model with information from investigation and diagnosis imagistic techniques, including simulation or comparative analysis. [Angheluta & al, 2008] Furthermore, the work will carry on with studies on the walls degradation (in Sasspol) or inspecting the restoration work from Red Jampa or Tsatsapuri. Also tools study for traditional craft research will be available, with added information regarding intervention strategy elaboration.

Alexander, A. & Catanese, A. 2007, Leh Old Town Conservation Project Ladakh, Indian Himalayas, e_conservation Magazine, INDEX ISSUE 6, September 2008, Portugal. An∙gheluta, L., Striber, J., Radvan, R., Gomoiu, I., Dragomir, V. & Simileanu, M. 2008, Early Fungal Contamination Tracking, International Journal of Systems Applications, Engineering & Development, Issue 4, Volume 2. George, L. Heritage, G.L. & Large, R.G.A. (editors) 2009, Laser scanning for the environmental sciences, John Wiley & Sons Ltd., Chennai, India. Guarneri, M., Bartolini, L., Ferri De Collibus, M., Fornetti, G., Francucci, M, Paglia, E., Ricci, R., Kolar, J., Nemec, I., Strlic, M., 2008, RGB-ITR application for cultural heritage: a practical case approached in the church of SS. Trinity In Hrastovljie, CHRESP: Cultural Heritage Research Meets Practice, Ljubljana, Slovenia. Kazhdan M., Bolitho M. & Hoppe, H. 2006, Poisson Surface Reconstruction, Proceedings of the fourth Eurographics symposium on Geometry processing, Cagliari, Italy. Nicolaescu, A. & Alexander, A. 2008. Red Maitreya Temple - Leh, Ladakh Mural Conservation Project, e_conservation Magazine, INDEX ISSUE 6, September 2008, Portugal. Remondino, F. & Menna, F. 2008, Image-based surface measurement for close-range heritage documentation, The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B5. Beijing, China. Shan, J. & Toth, C.K. (editors) 2008, Topographic laser ranging and scanning, CRC Press Taylor & Francis Group, Boca Raton, USA. Toubekis, G. Mayer, I. Döring-Williams, M. Maeda, K. Yamauchi, K. Taniguchi, Y. Morimoto, S. Petzet, M. Jarke, M. & Jansen, M. 2009, Preservation and management of the UNESCO World Heritage site of Bamiyan: laser scan documentation and virtual reconstruction of the destroyed Buddha figures and the archaeological remains 22nd CIPA Symposium, October 11–15, Kyoto, Japan. Visintini, D, Siotto E.B. & Menean E.A., 2009, 3D modeling of the St.Anthony Abbot Church in S. Daniele del Friuli (I):from laser scanning and photogrammetry to vrml/x3d model, Proceedings of the 3rd ISPRS International Workshop, Trento, Italy.

202

LACONA8.indb 202

11/10/2010 5:37:40 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Robotized structured light system for automated 3D documenting of cultural heritage R. Sitnik, M. Karaszewski, W. Załuski & P. Bolewicki Warsaw University of Technology, Warsaw, Poland

ABSTRACT: In this paper a fully automated 3D shape measurement system is presented. It consists of a rotary stage for placement of cultural heritage objects, a vertical linear stage with a mounted robot arm (with six degrees of freedom) and a structured light measurement setup attached to its head. All these manipulation devices are automatically controlled by a collision detection and a next-best-view calculation modules. The goal of the whole system is to automatically (without any user interaction) and rapidly (operation time decreased from days or weeks to hours) measure the whole object. The measurement head is automatically calibrated by the system and its possible working volume spans from centimeters up to one meter. Preliminary measurement results with assessment of hardware positioning accuracy and views integration algorithms are presented. 1

INTRODUCTION

In the present times, the digitization of shape of various types of items, either technical or industrial, as well as cultural heritage objects, is becoming more and more popular. The scanning devices used for this purpose are becoming more advanced and attain greater accuracy and resolution (Zhang et al. 2006, Kowarschik et al. 2000, Sitnik et al. 2005). Unfortunately, measurements of objects of complicated shapes, which are, at least among relics, very common, are time-consuming and laborintensive. Scanning and data processing have to be conducted by trained personnel. In general cases, the digitization process consists of taking multiple directional scans of an object (to obtain the complete representation of its surface), manual or semiautomatic data preprocessing (in most cases data is in the form of a cloud of points—a set of points forming the sampled object’s surface) and initial data stitching, which is mostly done manually by a qualified operator (Cyganek et al. 2009). The final stage of data processing is usually fine stitching of clouds, performed in most cases automatically using iterative algorithms (Ikeuchi et al. 2003). The need for a great number of manual operations, which can be done only by highly skilled personnel at the highest pitch, increases time and cost of digitization of a selected object. These shortcomings suppress the popularity of modern methods of shape documentation by 3D scanning. This is extremely important while dealing with cultural heritage objects, when the possibilities of application of their three dimensional models are numerous (Ikeuchi et al. 2007).

The only solution to the problems described is the complete automation of the whole digitization process. Because of this, authors of this paper developed a fully automated 3D shape measurement system, designed for cultural heritage objects scanning. The presented system does not require any interaction with operator apart from placing the object in the measurement volume, setting several input parameters and starting the measurement process. 2

CONCEPT

In the classic measurement the interference of an operator is required while taking directional scans, as well as during initial data stitching and in further processing. The presented system solves those problems by replacing operator’s actions with specialized algorithms and a robotic system for the manipulation of position and orientation of the measurement head (3D scanner) with respect to the measured object. 2.1

Best next measurement head position

In most cases, the digitization of a real object requires taking more than one directional measurement. This is caused either by object’s dimensions being greater than scanner’s measurement volume or by some part of object’s surface being blotted out by another element of the object. In these cases, another directional measurements have to be taken, each from different position of the measurement head. Placement of the scanner however is not a trivial problem, because

203

LACONA8.indb 203

11/10/2010 5:37:41 PM

while the number of directional scans should be as small as possible (because of the time required to capture and process the data), obtaining a full representation of the object surface with minimum data redundancy is hard to achieve. Moreover, there are no general rules to follow while dealing with various objects and finding good scanner locations is a matter of operator’s experience. In automatic measurements, the data is processed without user intaction and therefore minimum data redundancy is not so critical, though still important. There are two ways of selecting consecutive positions of the scanner—the first method is to take measurements from evenly sampled positions (for example rotate the object by 100°, take the scan, rotate by another 100° and so on) (Sitnik et al. 2002). Data redundancy in this case however is unnecessarily high, but even more important is the inability to deal with all cases of blotting out of some elements by other parts of the measured object. Another way for scanner placement is to use a specialized algorithm, which calculates the next position of measurement head basing on already collected data. This model is optimal, both from the point of view of data redundancy and dealing with blotted out parts of the object. The system which is presented in this paper calculates the required scanner position using this model. 2.2

Figure 2.

Photograph of the measurement head.

i.e. a rotating table for object placement, a vertical column with pedestal (both devices are driven by servomechanisms) and a commercial industrial 6-degrees-of-freedom robot fixed to the pedestal. The measurement device (in this case a structured light scanner) is mounted to the wrist of the robot (see Figure 2). The proposed configuration of devices gives many possibilities of scanner positioning, allowing to measure many types of objects. Application of a rotating table permits to scan the object from different directions. The linear vertical column makes it possible to measure high objects., while the 6-degrees-of-freedom robot provides the most precise positioning (translation and rotation) of the measurement head in relation to the measured object surface.

Measurement head positioning

Calculation of the best position of the scanner for the next measurement is not sufficient if one does not have means to place the measurement head in the selected point accurately. For a fully automated system, the scanner has to be positioned by some kind of a robotized system. The presented system (see Figure 1) uses three devices for this purpose,

2.3 Coarse and fine views integration After moving the scanner to a new position, next directional measurement is performed. The cloud of points obtained from this measurement does not contain any information about scanner position, therefore cannot be automatically transformed for initial fitting into a previously obtained dataset. In the automated system however, the position and orientation of the measurement head is known when the scan is performed, therefore this information can be used for cloud transformation. This procedure replaces initial data fitting, which is done manually or semi-manually by an operator in the classical digitization process (Pauly et al. 2005). After this initial data stitching, iterative fitting can be run. In the presented system, fine fitting is done by Iterative Closest Point algorithms (Chen et al. 1991). The newly obtained dataset is used as a source of data for calculation of the next scanner head position. The whole process continues in a loop until the whole measureable surface of the object is scanned. 3

Figure 1. Visualization of a robotized measurement system with local coordinate systems of each separate hardware module.

SYSTEM CALIBRATION

The crucial factor influencing the accuracy of manipulating of the measurement head is the

204

LACONA8.indb 204

11/10/2010 5:37:41 PM

calibration of all devices which are used for this purpose altogether as a framework. In this case, the term calibration refers to finding the relation between coordinate systems of devices and coordinate system of the scanner (SC). In the proposed system there are three independent devices (rotating table TC, column CA, and 6-DOF robot RC—see Figure 1). In consecutive sections, the calibration procedures of each of those devices are described. Although those procedures are different, finding transformation in each case is based on measurement of a calibration object from different scanner positions and fitting obtained clouds of points. The transformation matrices of fitted clouds are used for finding relations between coordinate systems of the calibrated device and the scanner. To calculate the coordinates of a point given in the Scanner Coordinate system (SC) in any other system, point’s original coordinates have to be multiplied by the homogenous transformation matrix (M) which transforms one coordinate system into another (Sitnik et al. 2009). To obtain the transformation from the secondary system to the original one, the transformation matrix has to be inverted. The presented system uses 3 transformation matrices—the first one binds (SC) and (RC) systems, the second one—(RC) with (TC) and the last one—(RC) with (CA). 3.1

Calibration of the robot coordinate system

The first stage of the calibration is the procedure of finding the relation between the coordinate systems of the robot (RC) and the scanner (SC), in which clouds of points are obtained. No relations between those systems are known, therefore at least six measurements have to be made to calculate them. The procedure of calibration is divided into two parts. Frst the scanner is translated repeatedly along a known axis by a known distance (in the RC system) and the measurements of objects are performed. This part of calibration is used for defining equations of versors of the RC system in Scanner Coordinates (SC). The second stage of calibration consists of rotation of the scanner head around a known RC axis to view object from different angles. The transformation matrices of clouds of points obtained with this step are used for calculation of the vector connecting the centers of RC and SC systems, the vector is defined in SC coordinates. From this moment each point defined in scanner coordinates can be transformed to the robot coordinate system and vice-versa. 3.2

Calibration of the rotary stage coordinate system

The initial relation between the (RC) and (TC) systems is unknown, because those devices can be

mounted in an arbitrary manner. The calibration procedure consists of taking numerous scans of the object placed on the table. Each scan is performed for different angular positions of the table. On the basis of obtained clouds’ transformations the axis of the rotation of the table is calculated. For simplicity it is assumed that the axis represents the Z axis of the (TC) system. Axis X is calculated as perpendicular to Z and intersecting the center of the (RC) system. The third axis, Y of (TC) is obtained from X and Z cross product. 3.3 Calibration of the axis of the vertical stage The kinematic chain representing all devices used for scanner manipulation is defined in a way which does not require full information about vertical column position. The only required parameter is the direction of the line along which the pedestal is shifted. To obtain the directional vector of this line (CA) it is enough to compare two clouds of points obtained with the scanner placed in two different positions of the pedestal. Those clouds are shifted along a line which is parallel to the sought line (CA). 4

NEXT BEST VIEW CALCULATION

When the first directional cloud of points is captured, the next best view algorithm is executed. It consists of two main parts: Rough Model Coverage (RMC) and Hole Filling (HF). The (RMC) algorithm is called first, and searches for possible directions of movement for the measurement system. If there are none, the (HF) algorithm starts and if it finds any directions, they are set as the output of the procedure, otherwise the measurement process stops. The (RMC) algorithm searches for borders of each directional cloud of points and, if they are not covered by any other directional clouds, these areas become bases of the next measurement positions calculation. First, (RMC) divides each directional cloud of points in its local coordinate space into nine areas with roughly equal number of points. The dimensions of these areas are defined by four planes parallel to the direction of measurement (see Figure 3a). For each area the algorithm calculates its coverage with other clouds of points. If at least 50% of the surface within the area is covered by other clouds of points, this area becomes inactive. Otherwise, a best fitting plane is calculated for this area. The center of the area and the corresponding normal vector are added to the set of next best views for the measurement. If the (RMC) returns an empty set of possible next views, the whole object surface is measured with exclusion of areas smaller than half of one ninth of

205

LACONA8.indb 205

11/10/2010 5:37:42 PM

Figure 4. Figure 3. Next best view calculation: a) RMC algorithm, b) HF algorithm.

the measurement volume size of the measurement head. The (HF) algorithm searches for holes in the multi-view model (see Figure 3b). The coordinates of the center and the best fitting plane formula are calculated for each hole and added as a possible next best view. Also in this stage all possible next measurement directions are compared. If two or more positions and orientations of the measurement head are similar (similarity means that identified holes can be captured during single measurement) then they are merged into a single next best view. 5

Digital model of goddess Kybele.

comparative measurement has been performed with manual positioning of the measurement head around the object by an experienced user and with interactive directional cloud of points integration. In this case the whole process took five working days (around 42 hours). Most of this time was spent on manual positioning of the system with respect to the object and interaction with the processing software. During tests the whole system was validated by comparative measurements of reference objects performed by the developed system and a portable measuring device Romer Sigma Arm 2022. The conclusion from these measurements is an initial assessment of measurement uncertainty equal to ± 40 μm.

EXEMPLARY RESULTS

The presented system was built with an industrial robot (Fanuc LR Mate 200i) with manipulation range of 700 mm, a linear column, which allows to shift vertically the robot mounted on the column pedestal. The robot can be ascended by a maximum value of 1800 mm. The third device used was a rotating table with the radius of 500 mm and 2000 kg maximum load. The measurement head used during measurements is a 3DMADMAC structured light scanner (Sitnik et al. 2005), but in general any scanner which returns results in the form of a cloud of points can be used. The accuracy of the scanning device is equal to 10−4 with respect to the measurement volume size, which can be defined as a prism of dimensions 170 mm × 200 mm × 100 mm. The test object presented below is a sculpture of goddess Kybele, belonging to the National Museum in Warsaw. Digitized version of this sculpture obtained with the presented system is shown in Figure 4. The whole automatic measurement of the sculpture and integration of directional clouds of points took less than four hours with an uncertainty less than 0,01 mm. The resulting directional cloud of points count is equal to 27 with more than 80M measurement points (see Figure 4). Also, a

6

CONCLUSIONS

In this paper an automated 3D scanning system for digitizing objects of cultural heritage is presented. Because of full automation of both measuring process and initial data stitching, the time required for digitization (measured from beginning of measurements to obtaining a full 3D model of an object) is significantly shorter than in classical scanning systems, in which many operations have to be done manually by qualified personnel. This improvement is crucial while dealing with digitization of numerous collections of relics because time and personnel required for measurements correspond in a direct way with the cost of digitization, and lowering the cost of those projects leads to their further popularization. The presented system is still in the development stage, the authors plan to augment it with automatic calibration procedures, because the current manual calibration process is time-consuming and laborintensive. Furthermore there are plans to develop a mobile system for the measurement of objects which from various reasons cannot be delivered to the measurement laboratory (for example elements

206

LACONA8.indb 206

11/10/2010 5:37:42 PM

of elevation of antique buildings). This system is planned to be semi-automatic, with manual positioning of the scanner head and its position and orientation being tracked by some, probably optical, system. ACKNOWLEDGEMENTS This work was performed under the grant No. R17 001 02 financed by the Polish Ministry of Science and Higher Education. REFERENCES Chen, Y. & Medioni, G. 1991. Object Modeling by Registration of Multiple Range Images: Proc. IEEE Conference on Robotics and Automation. 3: 2724–2729. Cyganek, B. & Siebert, J.P. 2009. An Introduction to 3D Computer Vision Techniques and Algorithms. Wiley, Sussex. Ikeuchi, K., Nakazawa, A., Hasegawa, K. & Ohishi, T. 2003. The Great Buddha Project: Modelling Cultural Heritage through Observation. Proc. The Second IEEE and ACM International Symposium on Mixed and Augmented Reality: 7–16.

Ikeuchi, K. & Miyazaki, D. 2007. Digitally Archiving Cultural Objects. Springer. Kowarschik, R., Kühmstedt, P., Gerber, J., Schreiber, W. & Notni, G. 2000. Adaptive optical three-dimensional measurement with structured light. Opt. Eng. 39: 150–158. Pauly, M., Mitra, N.J., Giesen, J., Gross, M. & Guibas, L.J. 2005. Example-Based 3D Scan Completion. Proc. Eurographics symposium on Geometry processing: 23. Sitnik, R., Kujawinska, M. & Woznicki, J. 2002. Digital fringe projection system for large-volume 360-deg shape measurement. Opt. Eng. 41: 443–449. Sitnik, R., Kujawińska, M. & Załuski W. 2005. 3DMADMAC system: optical 3D shape acquisition and processing path for VR applications. Proc. SPIE 5857: 106–117. Sitnik, R., Karaszewski, M., Zaluski, W. & Bolewicki, P. 2009. Automated full-3D shape measurement of cultural heritage objects. Proc. SPIE 7391: 73910K. Zhang, S. & Huang, P.S. 2006. High-resolution, Real-time 3-D Shape Measurement. Opt. Eng. 45(12): 123601.

207

LACONA8.indb 207

11/10/2010 5:37:43 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Through-glass structural examination of Hinterglasmalerei by Optical Coherence Tomography M. Iwanicka, L. Tymińska-Widmer & B.J. Rouba Institute for the Study, Restoration and Conservation of Cultural Heritage, Nicolaus Copernicus University, Toruń, Poland

E.A. Kwiatkowska, M. Sylwestrzak & P. Targowski Institute of Physics, Nicolaus Copernicus University, Toruń, Poland

ABSTRACT: Optical Coherence Tomography is non-contact and non-invasive technique of depthresolved imaging within media scattering and/or absorbing near-infrared light moderately. Its ability of remote sensing renders OCT especially convenient for examination of reverse painting on glass (Hinterglasmalerei). In this case, uniquely to other examination techniques, layers closest to the spectator are easiest to investigate. Moreover, the inspection is possible without dissembling the picture from the frame and protective back cover. In this contribution the OCT technique will be presented and results discussed with emphasis to its applicability to resolving specific conservation problems related to the reverse-glass paintings. Results obtained with an exemplary nineteenth century Hinterglasmalerei painting are discussed. 1 1.1

INTRODUCTION Overview

Reverse painting on glass (Hinterglasmalerei) is a painting technique of applying colours directly to the reverse side of the sheet of glass, which here becomes not only a support, but also a front protective (“varnish-like”) layer of the picture. On the contrary to conventional painting technique, the paint layers closest to the viewer, such as contour and glazes must be applied first, followed by opaque colours. Consequently, as a result of reversing of the whole painting process, the layers of a painting decisive to its aesthetics are directly adjacent to glass support, which makes them hardly accessible both for physical inspection and conservation treatment. Optical Coherence Tomography (OCT) is noncontact and non-invasive examining technique providing cross-sectional images of semi-transparent objects with micrometer resolution. It originates from medical diagnostics (Drexler et al. 2008) but it also has been successfully used in material science (Stifter 2007) and since 2003 for examination of various works of art (Targowski et al. 2004, Yang et al. 2004, Liang et al. 2005, Tymińska-Widmer et al. 2007, Liang et al. 2008, Targowski et al. 2008, Kunicki-Goldfinger et al. 2009). A complete list of papers on application of OCT in conservation one may find at www.oct4art.eu.

In this contribution utilisation of OCT as a tool to investigation of structure of reverse painting on glass is presented. Given that the supporting glass is still transparent, examination through its volume enables direct analysis of most exposed and important layers of the object, as well as phenomena occurring between support and paint, without necessity of dissembling the picture from its frame and backing paper. 1.2

Hinterglasmalerei—history and principle of the technique

The earliest known relics combining pieces of glass with painted decorations originate from antiquity. In the technique of reverse painting on glass, however, unlike in many others (like enamel or stained glass), the colours are not fired, but spread directly on the glass plate and intended to be viewed in backscattered light. The origin of Hinterglasmalerei and continuity of its development gives raise to scholarly controversy (Caldararo 1997). The major hypothesis are that it has been derived from Byzantium and rediscovered in Central Europe in late Middle Ages or that it had evolved in Europe from Roman times until the nineteenth century. In renaissance Italy reverse painting on glass became used in decorative arts as an imitation of enamel glass (Dupont et al. 1987). Along with the development of glass-making technology the technique

209

LACONA8.indb 209

11/10/2010 5:37:43 PM

of painting “behind glass” evolved and became utilised in making of miniatures and reproductions of great masters’ works. The latter phenomenon, having occurred mostly owing to growing popularity of engravings throughout Europe in the period of baroque (Steiner 2004), inevitably took effect in serial and increasingly routine bourgeois artistic production, especially popular in Bavaria and Austria in the eighteenth century. However, this was probably the way the Hinterglasmalerei technique had spread and became popular in folk art of South Germany and Central Europe from late eighteenth until middle twentieth century (Dupont and Hinrichs 1987, Błachowski 2004). Interesting example of such folk paintings are Romanian reverse-glass icons (Coman-Sipeanu et al. 2008). It is also worthwhile to note the twentieth century revival of Hinterglasmalerei in works of Paul Klee and some other German expressionists (BiglerGörtler 2003). An ethnographer Tadeusz Seweryn researched into the Polish Hinterglasmalerei painting tradition in the 1930s while it was still alive (Seweryn 1932). According to him, glass plates used in folk art production not only were of the most common kind, but also of inferior quality, bearing numerous flaws (air bubbles, drops, unevenness of thickness, wavy surface etc.) due to being production waste. The glass supports were likely to have been defatted and sized by an animal or fish glue. Then a composition was transferred often by means of a stencil or simply by redrawing an engraving (e.g. a xylograph). A contour and highlights must have been naturally painted first, followed by glazes and opaque paint layer. The background could have been painted overall, depending on the region of origin. Interestingly, using glass (smooth and nonabsorbent material) as a painting support, exerted an influence on a technique of applying of colours. It created the necessity to paint fast and without corrections. As a result, some artists used to alternate the painting media, employing ink to draw the contour, then oil colours for main composition, and finally, watercolours at background (Seweryn 1932). It must be emphasised, though, that this technique had a large number of variations. In general, the present condition of artefacts is usually poor. Inherent to the technique is a low adhesion of the paint layer to glass and detachments of whole paint layer is the typical conservation issue. Apart from that, discolouration of dyes and pigments occurs frequently. Due to the fact that this technique was the domain of folk craftsmen or experimenting modern artists, materials and particulars of artistic work vary significantly and are rarely predictable (Davison et al. 2003). For this reason, specific causes of deterioration may be difficult to reveal.

2 2.1

EXPERIMENTAL The OCT tomograph

The tomograms, shown in this report have been obtained with a prototype SOCT instrument based on an optical fibre Michelson interferometer set-up (Fig. 1), constructed in the Nicolaus Copernicus University. A broadband (Δλ = 107 nm, central wavelength 845 nm) superluminescent source (LS) has been employed. It generates radiation of high spatial (to ensure sensitivity) but low temporal (to ensure high axial resolution) coherence. The source is connected with a single mode 50:50 Fibre Coupler (FC) through an Optical Isolator (OI). After the coupler light is launched into two arms of the interferometer. Light propagating in a reference arm passes through a Polarization Controller (PC) used to provide the optimal conditions for interference, the Neutral Density Filter (NDF) for adjustment of the power of light to achieve the shot noise limited detection conditions and a block of glass acting as a Dispersion Compensator (DC). The light is then back reflected from the stationary Reference Mirror (RM) to the reference arm Fibre and Coupler (FC). The sample arm comprises galvo-scanners (X-Y) and lens. The light beam is scanned transversely across the object and backscatters and/or reflects from the elements of its structure and returns to the coupler FC. Light returning from the reference mirror and from the sample is brought to interference at the output of the interferometer and analyzed by a spectrometer comprising a volume phase holographic grating (DG) with 1200 lines/mm and achromatic lens.

Figure 1. Schematic of the SOCT instrument. LS – light source, OI – optical isolator, FC – fiber coupler, PC – polarization controller, NDF – neutral density filter, DC – dispersion corrector, RM – reference mirror, X-Y – transversal scanners, DG – diffraction grating, CCD – single line CCD camera, COMP – computer.

210

LACONA8.indb 210

11/10/2010 5:37:43 PM

The latter focuses the spectrum on a 12 bit line scan CCD camera (2048 pixels, 12 bit A/D conversion, Atmel). The spectral fringe pattern registered by this detector is then transferred to a personal computer (COMP). This signal after the Fourier transformation yields one line of the cross-sectional image (A-scan). It carries information about the locations of scattering centres in the object along the path of the penetrating beam. Transversal scanning across the sample enables collecting 2-D cross-section (B-scan). Additional scanning in the perpendicular direction gives 3-D information about a structure. In the instrument described, the axial resolution is equal to 4 μm (in air) and the lateral one is about 30 μm. Together with the instrumental limit of indepth imaging of about 2 mm all this make OCT tomography especially suitable for imaging layered structures like paintings for instance. The acquisition time for one A-scan is 40 μs and a thus whole cross-section (B-scan), usually composed of a few thousands of A-scans, is registered in a fraction of a second. Tomograms are usually shown as an intensity maps, coded in false-colour scale (Sylwestrzak et al. 2009) or like here in “reverse” gray scale: high scattering centres are shown as dark spots and low scattering media remains white or light gray. Since the axial resolution in OCT is usually much better than the lateral, tomograms are often shown with a in-depth axis expanded for better readability. 2.2

Figure 2. “Saint Wendelin”—nineteenth-century reverse painting on glass from Ethnographic Museum in Toruń, view in scattered light. Places of OCT examination are marked with white squares.

Object of examination

Complex issues regarding documentation and conservation of Hinterglasmalerei works can be illustrated by a nineteenth century depiction of Saint Wendelin, the painting of unknown technique and history of conservation (Fig. 2A). The picture, found in South Poland (Podhale region), and attributed to Slovak school had not been removed from the frame and backing paper in the last forty years since it entered the collection of the Ethnographic Museum in Toruń. The state of the painting is rather poor, given that it was under conservation at least twice. Two types of consolidation binders are present in the areas where the paint layer is missing or lifting from the support, as was revealed by inspection in UV/VIS. The overall adhesion of the paint layer to the glass was difficult to evaluate (Fig. 3) nonetheless, and dissembling the picture for inspection was not recommended owing to its poor condition. The OCT technique, however, as operating through the glass support, has offered the possibility to solve certain conservation questions concerning picture’s structure and the range of its damage.

Figure 3.

3

“Saint Wendelin” (fragment) in raking light.

RESULTS

The painting was analysed in twenty five locations by means of OCT (Fig. 2). The 3D data in a form of series of tomograms were collected from square areas of 7 to 12 mm width. Only the exemplary OCT tomograms are shown in Figures 4 and 5, but the conclusions are derived

211

LACONA8.indb 211

11/10/2010 5:37:43 PM

Figure 4. Fragment of the painting “Saint Wendelin” (right) showing the locations of the respective OCT tomograms (left, details in text).

Figure 5. Microphotography of the painting “Saint Wendelin” (bottom) showing the exact location of the respective OCT tomogram (top), arrows point to the corresponding areas (details in text).

from all the collected data. In the figures the bottom surface of glass (1 in Fig. 4a, b) is seen as the dark uppermost line, while the outer surface is not visible being out of the field of view. Below the glass, the structure of the painting may be recognised. The ability of imaging paint layers by means of OCT depends mostly on the absorption of near infrared light within the layer. This can be illustrated by the case of the area of contour (marked as 2a in Fig. 4a), which absorbs NIR completely and therefore only its upper interface is visible in the tomogram. The surrounding paint, however, seems to be transparent enough for the OCT examination, and, as can be resolved from the OCT images, consists of two layers (2b in Fig. 4a, b). Inspection of the boundary between the glass support and the paint reveals detachments at various stages of formation. Since the areas coded in black mark the highest scattering of light, they should be interpreted as the interfaces between media of contrasting refractive indices. This suggests the presence of air within the structure of the painting since different painting materials and glass have similar refractive index n ≈ 1.5 in contrary to air (n = 1). Apart from developed blisters or even flakes (4 in Fig. 4a, b), the thin layer of air (below in-depth resolution and thus seen as a

single dark line, 5 in Fig. 4b) indicates the areas of early delamination. Moreover, the OCT technique may assist in evaluation of the results of previous conservation treatment. In this case, the range and state of preservation of consolidation adhesive (3 in Fig. 4a, b) can be assessed. As it is visible in Fig. 4a, the flaking paint layer is actually submerged in the consolidant, which not only partially fills in the blisters, but also covers the back of the whole paint layer. Its bottom surface is clearly recognizable in the tomogram (Fig. 4a). It is present also in vast losses of paint layer (3 in Fig. 4b). It is cracked there and severely lacking adhesion to the glass, as can be seen in the central and the righthand side of the tomogram. In certain situations, the OCT technique may also aid in resolving issues concerning the cause of deterioration. Such is the case of partial discolouration of green paint layer within the background in “Saint Wendelin”, which seems to have turned dark and brownish in the lower part of the painting (Fig. 2). Direct comparison of the microphotography of exemplary discoloured area with the OCT tomogram, enables one to link this discolouration with the certain configuration of layers. Since the contours (2a in Fig. 5), as they absorb probing light completely, can be straightforwardly recognised in the OCT, the match between the macrophotography and the tomogram and hence localisation of discoloured areas on the latter can be done undoubtfully. The differences in the structure of the painting in its green and darkened brownish areas thus become clear. It is fairly visible that the unchanged paint layer in the left-hand side part of the tomogram consists of two paint strata of which the lower one is scattering and absorbing light considerably and the bottom surface of whole structure is not visible (2b in Fig. 5). On the contrary, in the discoloured region, there is only one layer of paint, detaching from the glass and covered with consolidation adhesive from both sides. Moreover, the bottom surface of the binder is visible in the tomogram (3 in Fig. 5) only in the region of discolouration. Since the whole paint layer is immersed in the adhesive (as was

212

LACONA8.indb 212

11/10/2010 5:37:44 PM

confirmed in other areas examined), this effect is attributable to the higher absorption of NIR in well preserved areas of the green. This is likely to have been caused by existence of another, more opaque layer underneath. Knowing that “Saint Wendelin” had been apparently created according to Slovak painting manner, which employed applying a white background overall the paint layer (Seweryn 1932), one may draw a conclusion that the discolouration occurred in areas which lack this kind of protection from the surrounding environment. Resolving of actual chemical processes is, however, still under investigation, and beyond the scope of this contribution. 4

CONCLUSION

In view of practical difficulties usually encountered during the conservation of reverse paintings on glass, resolving the range and progression of damage may be essential for daily handling of an artwork as well as planning of conservation treatment. As it was presented it is possible to identify even early stages of common delamination between glass support and paint layers with OCT. Especial advantage of using OCT for Hinterglasmalerei objects arrives from the possibility of examining through the supporting glass with direct access to the layers decisive for the visual effect. Moreover, Optical Coherence Tomography proves to be a useful tool for recognising stratigraphy of multilayered paintings and thus minimising classical sampling. Collecting material for such analysis endangers particularly this type of objects due to brittleness of the paint layer and its poor adhesion to the support. Additionally, since it is possible to create a map of detachments within the painting by means of OCT, this examination may aid in preventing accidental damage to the paint layer during sampling, by avoiding vast areas of very low adhesion. Moreover, by enabling the non-invasive identification of the locations of previous interventions, OCT assists in choosing representative areas for the sample collection. Furthermore, it is expected that OCT may be useful in recognition of possible structural deterioration of supporting glass. Apart from common mechanical damage, a curious instance may be delamination and inner cracking of sheet of glass owing to technological errors (Targowski et al. 2009). In conclusion, it must be emphasised that the OCT examination described here was performed without dissembling the Hinterglasmalerei picture, which is a great advantage while dealing with this kind of artworks. At this step there was no possibility of employing a chemical analysis (by neither invasive

nor non-invasive methods, since, for instance, the depth-resolved μXRF spectroscopy (Kanngießer et al. 2008) is not able to operate through glass). Therefore, the explanation of the chemical processes responsible for the damage revealed in the OCT tomograms should be considered as the next and complimentary stage of analysis. ACKNOWLEDGEMENTS This project is funded by Polish Government Research Grant trough years 2008–2011. EK, MS, and MI gratefully acknowledge additional support from European Social Fund and Polish Government within Integrated Regional Development Operational Programme, Action 2.6, by project “Stypendia dla doktorantów 2008/2008— ZPORR” of Kuyavian-Pomeranian Voivodeship. MI additionally acknowledges support from the project operated within the Foundation for Polish Science Ventures Programme financed by the EU European Regional Development Fund. Ethnographic Museum in Toruń is gratefully acknowledged for providing objects for examination. Authors wish also to thank Mr Waldemar Grzesik for professional assistance with photographs. REFERENCES Bigler-Görtler, J. 2003. Die Hinterglasmalerei von Paul Klee. Zeitschrift für Kunsttechnologie und Konservierung 17(1): 5–37. Błachowski, A. 2004. Malarstwo na szkle. Tradycje i współczesność polskiej sztuki ludowej. Lublin, Toruń, Polskie Towarzystwo Ludoznawcze – Oddział w Toruniu. Caldararo, N. 1997. Conservation treatment of paintings on ceramic and glass: two case studies. Studies in Conservation 42: 157–164. Coman-Sipeanu, O. & Guttmann, M. 2008. The Transylvanian glass icon collection of the ASTRA Museum, Sibiu, Romania: a conservation strategy. In J. Bridgland, (ed.) ICOM committee for conservation: 15th triennial conference, New Delhi, India, 22–26 September 2008: 590–595. Davison, S. & Newton, R. 2003. Conservation and Restoration of glass. Oxford, Butterworth-Heinemann. Drexler, W. & Fujimoto, J.G., (eds.) 2008. Optical Coherence Tomography: Technology And Applications. Biological And Medical Physics, Biomedical Engineering. Berlin Heidelberg New York, Springer-Verlag. Dupont, M. & Hinrichs, K. 1987. Hinterglasmalerei/ reverse painting on glass in 18th and 19th century Southern Germany – reconstruction of 6 “hinterglas” techniques. In K. Grimstad, (ed.) ICOM committee for conservation: 8th triennial meeting, Sydney, Australia, September 6–11, 1987, Sydney: 957–967. Getty Conservation Institute.

213

LACONA8.indb 213

11/10/2010 5:37:45 PM

Kunicki-Goldfinger, J., Targowski, P., Góra, M., Karaszkiewicz, P. & Dzierżanowski, P. 2009. Characterization of Glass Surface Morphology by Optical Coherence Tomography. Studies in Conservation 54: 117–128. Kanngießer, B., Mantouvalou, I., Malzer, W., Wolff, T. & Hahn, O. 2008. Non-destructive, depth resolved investigation of corrosion layers of historical glass objects by 3D Micro X-ray fluorescence analysis. Journal of Analytical Atomic Spectrometry 23: 814–819. Liang, H., Cid, M., Cucu, R., Dobre, G., Podoleanu, A., Pedro, J. & Saunders, D. 2005. En-face optical coherence tomography–a novel application of non-invasive imaging to art conservation. Optics Express 13(16): 6133–6144. Liang, H., Peric, B., Hughes, M., Podoleanu, A.G., Spring, M. & Roehrs, S. 2008. Optical Coherence Tomography in archaeological and conservation science – a new emerging field. Proceedings of SPIE 7139: 713915-713915-9. Seweryn, T. 1932. Technika malowania ludowych obrazków na szkle (La technique des peintures populaires sur verre) Lwów, with abstract in French. Steiner, W. 2004. Hinterglas und Kupferstich: 100 bisher unveröffentlichte Hinterglasgemälde und ihre Vorlagen aus drei Jahrhunderten (1550–1850). München, Hirmer Verlag. Stifter, D. 2007. Beyond biomedicine: a review of alternative applications and developments for optical coherence tomography. Applied Physics B Lasers and Optics 88(3): 337–357. Sylwestrzak, M., Kwiatkowska, E.A., Karaszkiewicz, P., Iwanicka, M. & Targowski, P. 2009. Application of graphically oriented programming to imaging

of structure deterioration of historic glass by Optical Coherence Tomography. Proceedings of SPIE 7391: 739109-1. www.oct4art.eu – a complete reference to articles devoted to application of OCT to examination of artworks. Targowski, P., Karaszkiewicz, P., Rouba, B.J., Markowski, D., Tymińska – Widmer, L., Iwanicka, M., Kwiatkowska, E.A. & Sylwestrzak, M. 2009. Optical Coherence Tomography for Non-invasive Investigation of Structure and Properties of Historic Glass. In The Art of Collaboration: Stained-Glass Conservation in the Twenty-First Century, CV US Conservation Studies I, New York: in press. Brepols Publishers. Targowski, P., Rouba, B., Góra, M., Tymińska-Widmer, L., Marczak, J. & Kowalczyk, A. 2008. Optical coherence tomography in art diagnostic and restoration. Applied Physics A: Materials Science and Processing 92: 1–9. Targowski, P., Rouba, B., Wojtkowski, M. & Kowalczyk, A. 2004. The application of optical coherence tomography to non-destructive examination of museum objects. Studies in Conservation 49(2): 107–114. Tymińska-Widmer, L., Targowski, P., Góra, M., Iwanicka, M., Łękawa-Wysłouch, T. & Rouba, B. 2007. Optical Coherence Tomography – a Novel Tool for the Examination of Oil Paintings. In J.H. Townsend, (ed.) Conservation Science, Milan, Italy, May 10–11, 2007, Milan, Italy: 175–182. Archetype Publications. Yang, M.L., Lu, C.W., Hsu, I.J. & Yang, C.C. 2004. The use of Optical Coherence Tomography for monitoring the subsurface morphologies of archaic jades. Archaeometry 46(2): 171–182.

214

LACONA8.indb 214

11/10/2010 5:37:45 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

Editing protocol for the digital mapping of related imagistic investigations L.M. Angheluta National Institute for Research and Development, Optoelectronics INOE 2000, Magurele, Romania

ABSTRACT: This paper proposes an editing protocol for artwork investigation & diagnosis workflow that will allow the art restorers to obtain, using photonic devices, a digital data model that includes a complex physical and chemical characterization, key information for conservation status appreciation. Based on the investigation-monitoring at certain parameters, a complex report will indicate the most delicate areas to potential conservation conditions dynamics, and is expected to offer a scientific prognosis depending on time and/or on possible restoration interventions. The proposed procedure is consisted in several stages. The study emphasizes two ways for obtaining the 3-D digital model: 3-D laser scanning and the depth from stereo method. In each case there are provided steps to follow as well with discussions upon advantages and disadvantages of each case. The final product is a digital 3-D replica of the real world investigated objects that can be viewed from any angle on a personal computer. This digital model is set in a virtual world environment that allows the viewer to see all the data from the investigation devices, right on the objects surfaces. For a better understanding of the purpose of this paper, we present the results obtained with the Hypogeum Painted Tomb from Constanta (4th century A.D.) following the proposed workflow steps. 1

INTRODUCTION

The advancements of photonic technologies in the last decade permitted emerging of new and better methods for compositional and morphological investigation of the materials. The latest researches emphasize the development of the non-contact and non-invasive, thus non-destructive, investigation techniques, especially for the cultural heritage artifacts in poor conservation states. Currently are known several non-contact and non-invasive investigation and material characterization techniques that can offer imagistic and spectral data about the shape and consistency of an object (Simileanu M. et al, 2008). In this paper we propose a procedure to obtain a unified model of simultaneous visualizing, on a digital replica of a tridimensional surface, of the multilayer data from more than one investigation technique. The present paper reports preliminary result of a national project focused on complex system for imaging techniques for investigation / diagnosis / restoration of the multilayer structures from historical monuments. This project do not aim just to establish protocols for multi-technique analysis of samples, following the type of object and material (stone, metal, paint, paper, polymers) or degradation issue (alterations, depositions, detachments),

but a protocol available for real objects from museums or galleries and for historical buildings, too. The best protocols are imagined to help prognosis studies and intervention simulations. 2

NON-CONTACT INVESTIGATION TECHNIQUES

For the purpose of this paper, we will present some results obtained during the digital 3-D reconstruction of the Hypogeum Painted Tomb from Constanta. Only two investigation techniques were employed at this stage: the 3-D laser scanning and the multispectral imaging. And of course digital photography was used to ensure the realistic texturing of the final 3-D model. 2.1 The 3-D laser scanning device The 3-D laser scanning device used for the purpose of this paper uses the technique of triangulation based on the relationship between the direction of the emitted laser beam the direction of the detected reflection, providing this way information about points on the object surface. It is called ‘triangulation’ because a triangle is formed between the laser head, the point on the surface of the object, and camera detection; knowing the sides of the triangle

215

LACONA8.indb 215

11/10/2010 5:37:45 PM

and the camera angle, that can be determined by observing the position of the point recorded in the camera’s field of view. Having provided the information about the shape and the size of the triangle, it can finally be obtained the triangle corner coordinates with the laser spot on the surface of the object. Compared to the ‘time-of-flight’ method, the triangulation technique has much greater precision to the order of micrometers, but has limited distance towards the object, about a few meters. It is especially useful for small museum objects and environments (Ene D. et al, 2008). Following the classical procedures of processing information from the scanning, it will be obtained digital 3-D shapes or objects, compatible with most of the 3-D graphics processing software. 2.2

The multispectral camera

Recording digital images in different spectrum domains, like ultraviolet, infrared or visible, is done with a multispectral camera that uses a CCD sensor and spectral filters to change the capturing modes. The device is portable and can be mounted on a tripod. This camera can capture images in the ultraviolet domain (UV fluorescence and UV reflectance), within two bandwidths of near infrared domain, the visible domain; it also can make false color images, combining one of the near infrared mode with the visible one. The near infrared captured images, offers the possibility to observe the initial drawings of a painting, but it also can provide information about the previous repairs of the surface. 3

PROPOSED INVESTIGATION PROCEDURE

The proposed procedure is consisted in several steps. Some of the stages are exemplified with images and screens from the process of digital reconstruction of the Hypogeum Painted Tomb, from Constanta. This workflow has two branches, depending on the technology used to record the 3-D shape of the investigated surface. On one side we the 3-D laser scanning and on the other side the depth gaining from 2-D image pairs. 3.1

Investigation parameters and spatial limits calculations

At this starting level the investigators are preparing calculation sheets for the relation between the photonic devices’ image resolutions (in pixels) and the real world scene, at variable distances.

Figure 1.

Proposed procedure work flow diagram.

The pixel/mm2 ratios will help in the following steps for finding the stereo correspondence or for texturing. 3.2

Decision: 3-D laser scanner or 3-D from stereo (or both)

At this point most of the investigators choose depending on their own resources either to use a 3-D laser scanner or a stereo configuration of two cameras (or one horizontal sliding camera) in order to obtain the digital 3-D shape of the real world investigated scene. While with the 3-D laser scanner it is easier to obtain a 3-D model of the real world scene, it could be quite expensive for some investigators. A stereo configuration is much cheaper, but involves a lot of complex post-processing procedures. But in a stereo configuration there can be used all sort of photonic investigation devices that provides results as 2-D images. Therefore it can be used with multispectral cameras, therma-cameras, usual digital camera, etc. 3.3

3-D laser scanning

After setting the conditions established in the first steps, 3-D laser scans are performed, and the digital 3-D objects are processed and the three dimensional model is obtained. The model is not textured. At this step, if the investigated object or surface is too big to fit in a single scan, there are performed

216

LACONA8.indb 216

11/10/2010 5:37:45 PM

Figure 3. Example of a tomb’s vaulted ceiling digital mesh map.

Figure 2. Right-hand wall and vaulted ceiling 3-D scan of the tomb.

surface. Therefore attention must be provided when investigating large surfaces, in order to have good matches between the areas covered by single images. 3.6 3-D mesh preparing

several scans and then the resulted objects are merged in order to obtain the full digital 3-D scene. The result is stored in several digital file formats that are supported by most of the 3-D modeling software. 3.4

The stereo configuration set-up

It is recommended a canonical stereo configuration, in order to obtain rectified images. This way the horizontal line of the pixels in both images coincides with the epi-polar line of the image planes. This will ease the correspondence search. A canonical stereo configuration has two constraints: the optical axes of the cameras must be parallel and the image planes must be in the same plane (I. Cox et al, 1996). The cameras used can be: usual digital camera, multispectral camera, therma-camera, etc. Actually any kind of investigation device that delivers results as 2-D images (even intensity distribution maps) can be used in this set-up. But both pair images must be taken with the same camera. 3.5

Investigations

At this stage the investigation devices records the necessary data of the real world scene. If the data are used for texturing the 3-D laser scanned model, they must be obtained from the same position as the scanner. In the case of the stereo recordings, the devices’ positions are shifted constantly until there is enough data to obtain a full digital replica of the 3-D scene. The images obtain must be recorded with identical conditions, and should cover the entire studied

Once the 3-D laser scanning is finished, and the parts are ensembled, the 3-D digital surface is processed so that it is obtained a 2-D mesh map of the object. This map will be used to texture the 2-D image data from the investigation devices on the 3-D digital model. As it can be seen in Figure 3, the vaulted ceiling of an underground tomb map is a 2-D flat image that has specific visual features. These features are used in the texturing of the surface, to match the images (investigation results) with the 3-D digital model. 3.7 2-D image pairs The image pairs are recorded two by two. Either there are used two identical investigation devices, to get each image, either it is employed a camera position shifting configuration 2. 3.8 Disparity map For each pair of images there will be a disparity map. This is actually an intensity distribution map of the disparities in the two images regarding the real world scene. The higher peaks, the greater or smaller the difference between the corresponding pixels (depending on the algorithms employed). In order to obtain these disparity maps, it must be solved the correspondence problem. 3.9 3-D model Using the disparity maps and by applying simple calculus there can be obtained a 3-D digital model

217

LACONA8.indb 217

11/10/2010 5:37:45 PM

of the investigated scene. The resolution of the model is directly linked to the resolution of the composing images. The result is a 3-D matrix, with the coordinates and depth of each pixel. This matrix is converted in a compatible 3-D file format. 3.10

Texturing

For the 3-D laser scanned mesh there are several algorithms of texturing using the mesh map from step 6. As for the 3-D object obtain from stereo image pairs, texturing can be done directly, using the merged coordinates of the two pairs. This way it is easily to texture the occluded areas, in some cases. If the investigate surface is large, it might require several images to cover the whole area. Therefore, the investigation areas must match perfectly, in order to merge the resulted images in a one big texture image. As an example, in Figure 4 is depicted the image texture for the vaulted ceiling of the painted tomb from Constanta. This image is constructed by merging 10 photos. Using the map features provided at step 6, these images are matched with the 3-D surface of the ceiling. 3.11

Final 3-D scene

The final three-dimensional model is to be included in a dedicated digital interactive virtual environment. This environment allows the user to visualize in different angles the investigated scene. More than that, it allows the study of the object by observing all the collected data in real-time on the digital model. This way there can be easily detected areas with different problems. This final product is either printed with a 3-D printer, or published on-line via a website, with a virtual world engine, that allows users from different parts of the world to access and visualize

Figure 4. Vaulted ceiling texture using digital photographs.

the investigation results mapped on the 3-D digital shape of the object. 4

PARTIAL RESULTS

This protocol was used to re-create digitally the Hypogeum Painted Tomb from Constanta, Romania. This is an IVth century AD underground tomb that was found several decades ago. The interior of the tomb was scanned with the 3-D laser scanner. Because the walls were flat the use of a stereo configuration was pointless. At this stage only digital photography and multispectral imaging studies were performed. The resulted 3-D model is embedded in an html web page with the VRML programming language. Using a virtual world environment the user can visualize the 3-D model of the investigated object, from any desired angle. In this way it is avoided the physical manipulation of the real object, that increases the chances of it being damaged. More than that, if you can manage to publish the results on the WWW, anyone who might be interested in studying that object (be it scientist, art restorer, historian, etc.) can have access to it anywhere in the world. This is an extremely useful tool for the art objects or archaeological sites that are hard to reach and study. Additional investigations were made with the multispectral camera to study the underlayers of the wall paintings. The virtual world environment allows the creators to define specific ‘sensitive’ areas, where the user can click and find out more specific information regarding those areas or results from different investigation devices.

Figure 5. Web-based interactive interface presenting the Hypogeum Painted Tomb from Constanta—full textured.

218

LACONA8.indb 218

11/10/2010 5:37:46 PM

d. History of the investigated artwork restorations, by highlighting any touch-ups, existing or concealed detachments in substrates; The information is, this way, available without the need to move the concerned expert on the spot. So he can access and view the digital model from his lab or from home, using a computer, and more importantly, the future studies on the object would be done without the need to handle physically the object. ACKNOWLEDGEMENT Figure 6. Screenshot with NIR underlayer detail of the painted bird area selection.

The full version of the Hypogeum Painted Tomb can be found at http://inoe.inoe.ro/IMAGIST/ resurse/hypogeu/page/run.htm 5

DISCUSSION

This study has been undertaken under the auspicies of the national research programs PNCDI II— Imagist—91009 and Program Nucleu. I would also like to thank eng. Dragos Ene and dr. eng. Roxana Radvan for their kindest collaboration and support during on-site measurements. REFERENCES

A first advantage of this method is that it uses non-contact and non-destructive high precision photonic techniques, to collect important data on the quality and conservation status of the surface and the hidden layers of an artifact, which can be fragile to mechanical contacts of any kind. A direct advantage of the digital model obtained this is way, is that it offers information very useful to the experts in the field of conservation-restoration of artworks, historians, insurers, curators. In short, the resulted model contains high precision data about the: a. Integrity, quality and the surface topography of the object b. Conservation status and possible biological attacks, with the mapping of their intensity distribution on the investigated surface c. Difference between compositions of pigments that may belong to interventions following the original paintings

Cox I., Hingorani S. and Rao S. “A Maximum Likelihood Stereo Algorithm”, Computer Vision and Image Understanding, Vol. 63, No. 3, May 1996. Ene D., Maracineanu W., Deciu C. and Radvan R. “Three dimensional imaging of cultural heritage as a basis for a knowledge cultural assets”, University Politehnica of Bucharest Scientific Bulletin-Series A-Applied Mathematics and Physics—Volume: 70, Issue: 2, Pages: 71–81, 2008. Sabry el-Hakim, Lorenzo Gonzo, Francesca Voltolini, Stefano Girardi, Alessandro Rizzi, Emily Whiting, “Detailed 3-D modeling of castles”, International Journal of Architectural Computing, issue 02. Vol. 05. Simileanu M., Maracineanu W., Striber J., Deciu C., Ene D., Angheluta L., Radvan R. and Savastru R. Advanced research technology for art and archaeology— ART4 ART mobile laboratory -JOAM. Vol. 10, nr. 2, pg. 470–473, 2008. Pollefeys M. and Van Gool L. Visual modeling: from images to images, The Journal of Visualization and Computer Animation, 13: 199–209, 2002.

219

LACONA8.indb 219

11/10/2010 5:37:47 PM

Lasers in the Conservation of Artworks VIII – Radvan et al. (eds) © 2011 Taylor & Francis Group, London, ISBN 978-0-415-58073-1

U-ITR: A 3D laser scanner prototype aimed at underwater archaeology applications R. Ricci, L. De Dominicis, M.F. De Collibus, G. Fornetti, M. Guarneri & M. Nuvoli ENEA, Frascati, Rome, Italy

M. Francucci ENEA fellow, Frascati, Rome, Italy

ABSTRACT: The U-ITR (Underwater Imaging Topological Radar) is a 3D laser scanner designed to operate directly in underwater environments and aimed at marine archaeological applications. A prototype is under advanced development at the ENEA Artificial Vision Laboratory (Frascati, Rome). The system is based on Amplitude Modulation (AM) rangefinding. This enables one to acquire in a single scan both range (distance) and reflectivity (imaging) information, and produce accurate, photorealistic, three-dimensional (3D) images. Preliminary results show that using a blue (405 nm), continuous-wave AM laser probe at a sufficiently high modulation frequency significantly reduces the optical noise due to water backscattering, resulting in better images and more accurate phase—i.e. distance—determination. Further improvements can be obtained by using conveniently arranged polarizing filters—yet at the cost of a loss of signal. 1

INTRODUCTION

Underwater 3D laser scanning is a mostly unexplored research field with potential applications in areas such as marine archaeology and inspection of submerged structures for industrial or scientific purposes. The task is challenging because, in natural waters, absorption and scattering of light by hydrosols and suspended particulates can spoil significantly the quality of 3D images. A few years ago the ENEA Artificial Vision laboratory started the realisation of the U-ITR, an underwater AM 3D laser scanner (Mullen et al. 2004, Bartolini et al. 2005) for marine archaeology. The system is now under advanced development and a first prototype will be operative in late 2010. The U-ITR uses a bistatic setup (Strand 1995) and a blue laser source at 405 nm, falling in region where absorption of clean seawater has a minimum. Even in relatively clean seawater, though, radiation backscattered by the medium gives rise to an unwanted and disturbing signal (optical noise) that strongly degrades the signal-to-optical-noise ratio S/N and contrast C attainable by the U-ITR. The negative effects of the optical noise can be minimized by: 1. using polarized laser radiation and a polarization selective detection scheme, i.e. by exploiting a proper polarimetric technique (Bartolini et al. 2007a, Bartolini et al. 2007b, Mullen et al. 2009);

2. selecting an optimal modulation frequency beyond a certain “cut-off ” frequency, depending on both the optical properties of the medium and the geometrical characteristics of the system (Pellen et al. 2000, Mullen et al. 2002, Mullen et al. 2004, Bartolini et al. 2008). We report the results of a series of laboratory experiments, aimed at verifying if and how the adoption of these methods helps increase the accuracy of the U-ITR and improve its performances, especially the quality of 3D images. 2

RESULTS

The experimental apparatus (Figure 1) consists of a test tank—equipped with an anti-reflection coated entrance optical window and having a

Figure 1.

Scheme of the experimental setup.

221

LACONA8.indb 221

11/10/2010 5:37:47 PM

length of either 1.5 m or 25 m, depending on the experiment—and an U-ITR laboratory prototype. The latter comprises a launching stage (AM diode laser emitting continuous-wave light at 405 nm with power of 20 mW; single-mode connection optical fiber with core diameter of 4 μm; launching optics) and a receiving stage (receiving optics; multi-mode connection optical fiber with core diameter of 1 mm; photomultiplier; lock-in amplifier). Backscattered light is collected by a short-focal-length lens and focused onto the photomultiplier (PMT, Hamamatsu H5783). The Stanford Research SR844 lock-in amplifier, beside measuring the amplitude and phase shift of the detected signal, is also used to modulate the laser intensity at frequencies up to 200 MHz. Light source and receiver are arranged in a bistatic layout for obtaining a partial spatial filtering of the optical noise by limiting the transmitter and receiver common field of view. A first experiment, exploiting a polarimetric technique, was performed on a white-painted, slightly bent metallic disc with diameter of 75 mm. The disc had a large hole in the center (diameter 33 mm), and four small holes equally spaced along the perimeter (diameter 4 mm). The target was located at a distance of 1.5 m from the receiver in a mixture obtained by adding 0.1 ml of Maalox® to 27 liters of tap water in order to reproduce a turbid, optically thin (τ = kz < 10, k = extinction coefficient, z = medium column length), scattering medium. The scanned area (40 × 80 pixels, sampling time for data element = 125 ms) covers the entire disc also including regions where only water was present, so enabling a comparison of the contrast between target and the surrounding medium. The results are shown in Figure 2 where two 3D images of the target are reported, each in two different orientations. The images were obtained by sweeping the target with Vertical (V) linearly polarized incident laser light and capturing only the component with linear polarization of the backscattered radiation, i.e. the component with vertical or Horizontal (H) polarization of the retro-diffused light. In other words, the images of Figure 2 are those obtained in copolarized (VV) and cross-polarized (VH) configurations, respectively. By observing Figure 2, it is possible to conclude that, for an optically thin medium and linearly polarized incident light, better imaging results are obtained in the VH scheme, due to the different depolarization properties of water and target. Specifically, the use of a cross-polarized working scheme enables one to achieve a more effective optical noise suppression with respect to the copolarized configuration and, so, to improve the 3D imaging performances of the system, in terms of both phase and contrast. In fact, the target image

Figure 2. 3D images of a white-painted metallic disc in turbid water (27 l of tap-water + 0.1 ml of Maalox®) located at distance of 1.5 m from receiver. Measurements were made by working at a laser modulation frequency of fm = 39 MHZ.

in VH polarizer configuration appears more clear and bright and the holes are better visible (Figure 2 on the bottom right) and shows the slightly curvature of target related to a more accurate range measurement (Figure 2 on the bottom left). On the contrary, the image obtained with the VV polarizer configuration appears blurred, the four small holes tend to disappear and the slightly curvature on target is no more observable because of a worse optical noise rejection (Figure 2 on the top). A second experiment was performed in similar conditions, but on a different target represented by a dark-gray-painted, sanded, metallic ladder. The latter had 1 cm-high steps, apart from the first step whose height was 4 cm. Two series of measurements were carried out by using respectively: 1) tap water (k = 0.06 m−1) and 2) a mixture of tap water and skim milk (k = 2 m−1). However, also in this case, the medium can be considered optically thin. In each scan, 40 × 80 arrays of data were acquired by sweeping the laser probe (fm = 39 MHz) perpendicularly onto the target, with a sampling time of 125 ms. Horizontal and vertical scanning angles were equal to 3.81° and 1.53° respectively, corresponding to a scan area of 10 cm (length) × 4 cm (height). The main results of this experiment are reported in Figure 3 and Figure 4, that show 3D images and an overplot of 40 linear phase profiles of the ladder, respectively. Specifically, 3D images recorded in the cross-polarized (VH) linear working scheme (Figure 3 on top and bottom left) evidence better phase measurement accuracy, contrast, spatial resolution (of the order of millimeter at 1.5 m), as well as less phase noise compared to 3D models recorded by using the copolarized (VV) linear

222

LACONA8.indb 222

11/10/2010 5:37:48 PM

Figure 3. 3D images of a dark-grey sanded metallic ladder immersed in water of varying total attenuation coefficients k, obtained by using proper polarimetric technique. The ladder-receiver distance is 1.5 m and the laser modulation frequency is fixed at fm = 39 MHz.

accurate determination of the step heights, with an estimated uncertainty in the measurement of distance of 0.5 mm at 1.5 m. A similar, yet slightly worse result was also obtained, always in clean water, in copolarized (VV) configuration (Figure 4 on top right). This occurs because the scattering phenomenon is almost negligible in clean water producing a minor effect on phase measurements. Thus, in this specific case, the use of polarizers in cross- or copolarized scheme does not significantly change the results. The situation is completely different in the case of turbid water (Figure 4 on bottom). In fact, in conditions of higher scattering rates, the accuracy of phase measurements remains acceptable, and the structure of the ladder is still well distinguishable, only in VH configuration (Figure 4 on bottom left), with an estimated uncertainty of 2 mm at 1.5 m. On the contrary, a significant degradation is evident in the linear phase profiles obtained by using the copolarized scheme (Figure 4 on bottom right), with distance uncertainty of 1 cm at 1.5 m. In summary, also these results confirm that, at least for an optically thin medium, more effective optical noise rejection is achieved both in clean and turbid waters by using a cross-polarized (VH) rather than a copolarized (VV) detection scheme. In order to verify the interference-like effect between water and target backscattering signals ( VˆW and VˆT , respectively, in complex notation) in an AM 3D laser imager as a function of the modulation frequency fm (Mullen et al. 2004, Bartolini et al. 2008), we report in Figure 5 the results of

Figure 4. Phase profiles of the ladder corresponding to the images reported in Figure 3. Relative distances determined by phase measurements in the VH scheme match well with real step separations—an indirect indication of the reduced incidence of phase disturbances due to optical noise.

working configuration (Figure 3 on top and bottom right). In the latter case, the steps look rougher and the ladder structure is smoothed due to the higher contribution of optical noise, especially for higher turbidity (k = 2 m−1). The theoretical phase differences (corresponding to the height of the ladder steps) can be calculated by means of the formula Δφ theor = 4π fm d/v, where v is the light speed in the medium and d the considered distance (i.e. target range). So, for the first step (d = 4 cm) one has Δφ theor ≅ 5° at 39 MHz, while for the others (d = 1 cm) Δφ theor ≅ 1.25°. From Figure 4 on top left, it is clear that the phases measured in cross-polarized (VH) configuration and in clean water correspond fairly well to the expected values, thus permitting an

Figure 5. Signal-to-optical-noise ratio as a function of fm for different values of k = kwater and zT (target range). Filled squares represent measurement results. Open squares are obtained after removing the optical noise— measured directly in a 25 m long tank—which is responsible of an interference-like effect.

223

LACONA8.indb 223

11/10/2010 5:37:48 PM

three different experiments, characterized by the following values of the parameters: 1. k = 0.66 m−1 and zT = 3.5 m (η = 0.08 on average, Figure 5 on top); 2. k = 0.88 m−1 and ZT = 3.5 m (η = 1.5 on average, Figure 5 on center); 3. k = 0.66 m−1 and ZT = 5.25 m (η = 0.6 on average, Figure 5 on bottom). In the adopted notation, ZT represents the target distance along the beam propagation direction, while η = VW/VT, where VW and VT denote the amplitudes of water and target backscattering signals, respectively. In all cases, the experimental conditions corresponded to optically thin medium in a transition regime between single and multiple scattering events being the optical thickness kzT < 10 (Berrocal et al. 2007). The water signal VˆW and the detected total signal Vˆ VˆW + VˆT , given by the complex sum of water and target backscattering contributions, were measured independently, at modulation frequencies fm varying in the range (0.5−146) MHz. In particular, the measurements of the water signal VˆW were directly carried out by shooting the laser beam in a black-walled 25 m long test tank with no interposed target, after verifying that the beam was fully attenuated before reaching the tank bottom. The following formulas were used for determining the signal-to-optical noise ratio R and the relevant quantity R′: R = V/Vw;

R′ = VT/VW ≡ 1/η

where VT = V(1 + η′2 – 2η′ cos ΔΦ′)1/2. Here V is the amplitude of the detected total signal, ΔΦ′ the phase difference between the detected and water signals and η′ = VW/V. It is worth noticing that, although Vˆ and VˆW were not measured in identical experimental conditions, VˆW represents a good approximation of the optical noise component of Vˆ for sufficiently large ZT, since the optical noise signal is dominated by light backscattered in the first few meters of the water column (Mullen et al. 2009) and, consequently, such a backscatter volume is equivalent to the infinite water column in terms of backscattered signal. From Figure 5, where the trends of R and R′versus fm for the three aforecited experiments are shown, one observes that the signal-to-optical-noise ratio R oscillates—with maxima and minima lying at specific values of fm depending on k, zT and spatial separation between light source and receiver— when water and target signals are comparable (η ≅ 1, Figure 5 on center and bottom), while it increases almost monotonically if η << 1 or η >> 1, i.e. if one of two component signals is clearly prevailing (Figure 5 on top).

The observed oscillations in R represent an experimental evidence of the existence of an interference-like effect between water and target backscattering signals in amplitude-modulated CW laser systems as U-ITR. Although not explicitly shown in this paper, similar considerations are also valid for the contrast C and for the amplitude V of the detected total signal as functions of fm (Mullen et al. 2004, Bartolini et al. 2008). It is also evident that R grows by increasing fm because of the low-pass filter frequency response of water backscattering. This constitutes an important verification of the general tendency of the U-ITR system to improve its performances as fm increases. Oscillations are instead drastically reduced for the (calculated) quantity R′ whenever η ≅ 1 (Figure 5 on center and bottom), with no significant effect in the other cases (η << 1 or η >> 1, Figure 5 on top). This occurs through the analytical removal of the water contribution from the detected total signal, that permits to calculate the target signal amplitude VT and, consequently, the non-oscillating quantity R′ after measuring VW, V and ΔΦ′. In summary these results, beside constituting a experimental proof concerning the interferencelike effect in amplitude-modulated laser systems (Mullen et al. 2004), show—as firstly suggested in Mullen et al. 2006—the possibility to improve the performances of an underwater amplitudemodulated 3D imager by working at modulation frequencies opportunely selected corresponding to the local maxima of R, even below the water cutoff frequency. On the other hand, even when the exact position of the local maxima is not known a priori, it is still possible to reduce the deleterious effects of the optical noise by direct cancellation (i.e. by calculating VT), provided the water backscattering signal can be measured independently by shooting in open water. Finally, a last experiment was devoted to the direct measurements of the water backscattering signal as a function of fm, in order to experimentally verify the low-pass filter frequency behavior of the optical noise in amplitude-modulated laser imager (Mullen et al. 1995, Pellen et al. 2000, Mullen et al. 2002). We used the experimental setup of Figure 1 with a transmitter-receiver separation of rrec = 0.2 m, a receiver field of view of θFOV ≅ 6.7° = 0.117 rad and a receiver radius of r0 = 0.025 m. The AM laser beam was shot in a fixed direction onto the blackwalled 25 m long test tank filled with water without interposed target, and the retro-diffused radiation measured by varying fm in the range 0.5 ÷ 138 MHz

224

LACONA8.indb 224

11/10/2010 5:37:50 PM

after verifying that the laser light was completely attenuated at the tank bottom. The extinction coefficient of turbid water used in the experiment, measured by using a PerkinElmer Lambda 25 UV/vis spectrometer, was equal to k = 0.66 m−1. The obtained experimental results, normalized with respect to the maximum value, are reported by means of filled squares in Figure 6, where a comparison with the theoretical outcomes (solid line) expected for an amplitude-modulated laser system (Ricci et al. in press) is also shown. The theoretical curve was calculated in the validity conditions of Small Angle Diffusion Approximation (SADA) (Mullen et al. 2002, Ricci et al., in press) and of Multi-Component Approach (Mullen et al. 2002) by neglecting absorption and laser beam spread due to multiple scattering events in the forward direction in the simplistic hypothesis that single scattering in the backward direction is the only attenuation mechanism. In this framework, the model parameters used for obtaining the theoretical curve are the following: 1 k ′ k = ksd 0 , k sf = 0 m −1 , θFOV = 0.117 rad, rrec = 0.2 m and r0 = 0.025 m.

Here k ′ k − ksf is the reduced extinction coefficient, ksf a f ks and ksd ad ks the forward and the diffuse scattering coefficients, respectively, while ks is the scattering coefficient with ks ksf + ksd . The dimensionless parameters af and ad are defined in the range [0, 1] with af + ad = 1 and, usually, af >> ad in the assumption of a scattering phase function strongly peaked in the forward direction, as generally occurs in natural waters as a consequence of scattering events from suspended large particles typical of intermediate or Mie regimes. The results reported in Figure 6 confirm both experimentally and theoretically the low-pass filter trend of water backscattering signal by varying fm, and clearly show that an effective optical noise rejection and, consequently, a performance improving (signal-to-optical noise ratio, contrast, range measurement accuracy) can be achieved in amplitude-modulated laser systems by increasing the laser modulation frequency beyond the cut-off frequency. The latter, defined as the frequency at which the power is 1 2 of the maximum value, is equal to fc = 26.09 MHz if calculated by means of the model and to fcexp = ( ± ) MHz if experimentally estimated in the specific case of the data of Figure 6. The two values of the cut-off frequency are in fairly good agreement, showing that the theoretical model provides a satisfactory description of the water backscattering low-pass filter behavior

Figure 6. Optical noise behavior vs modulation frequency for k = 0.66 m−1. Filled squares correspond to experimental measurements. The solid line is the theoretical expectation, calculated by neglecting absorption and laser beam spread.

as a function of fm in amplitude-modulated laser systems. Finally, a further improvement of the agreement between theory and experiment can be achieved by taking into account the absorption and the laser beam spread in the theoretical model. 3

CONCLUSIONS

The results reported in this paper show that the quality of underwater 3D images recorded by U-ITR, as well as its performances (signal-to-optical-noise ratio, contrast, range measurement accuracy), can be considerably improved: 1. by using a suitable polarimetric technique; 2. by operating at a modulation frequency either greater than the medium cut-off value or corresponding to signal-to-optical-noise ratio (or contrast) local maxima. Since part of the received power is lost on the polarizers when polarimetry is used, the second method is to be preferred in case of low detected signals. The oscillations of the signal-to-optical-noise ratio R and contrast C, observed in amplitudemodulated laser systems as U-ITR, are due to the interference-like overlapping of the target and water backscattering signals. The direct measurement of the water disturbance signal (optical noise), if possible, enables one to remove the corresponding contribution, yielding more accurate results with no oscillatory dependence on the modulation frequency, as shown by the calculation of the quantity R′ by varying fm.

225

LACONA8.indb 225

11/10/2010 5:37:51 PM

ACKNOWLEDGEMENTS Funding for this project was provided by the Italian national project “BLU-Archeosys”—Diagnostica per l’archeologia subacquea” (Diagnostics for underwater archaeology). REFERENCES Bartolini, L. De Dominicis, L. Ferri de Collibus, M. Fornetti, G. Guarneri, M. Paglia, E. Poggi, C. & Ricci, R. 2005. Underwater three-dimensional imaging with an amplitude-modulated laser radar at a 405 nm wavelength. Appl. Opt. 44: 7130–7135. Bartolini, L. De Dominicis, L. Ferri de Collibus, M. Fornetti, G. Francucci, M. Guarneri, M. Paglia, E. Poggi, C. & Ricci, R. 2007a. Polarimetry as tool to improve phase measurement in an amplitude modulated laser for submarine archaeological sites inspection. Proc. of SPIE 6618: 66180I-1–66180I-12. Bartolini, L. De Dominicis, L. Ferri de Collibus, M. Fornetti, G. Francucci, M. Guarneri, M. Nuvoli, M. Paglia, E. & Ricci, R. 2008. Experimental evidence of signal-optical noise interferencelike effect in underwater amplitude-modulated laser optical radar systems. Opt. Lett. 33: 2584–2586. Bartolini, L. De Dominicis, L. Fornetti, G. Francucci, M. Guarneri, M. Poggi, C. & Ricci, R. 2007b. Improvement in underwater phase measurement of an amplitude-modulated laser beam by polarimetric techniques. Opt. Lett. 32: 1402–1404. Berrocal, E. Sedarsky, D.L. Paciaroni, M.E. Meglinski, I.V. & Linne, M.A. 2007. Laser light scattering in turbid media Part I: “Experimental and simulated results for the spatial intensity distribution”. Opt. Express 15: 10649–10665.

Mullen, L.J. Zege, E.P. Katsev, I.L. & Prikhach, A.S. 2002. Modulated Lidar System: Experiment vs. Theory. Proc. of SPIE 4488: 25–35. Mullen, L. Vieira, A.J.C. Herczfeld, P.R. & Contarino, V.M. 1995. Application of RADAR Technology to Aerial LIDAR Systems for Enhancement of Shallow Underwater Target Detection. IEEE Transactions on Microwave Theory and Techniques 43: 2370–2377. Mullen, L. Laux, A. Concannon, B. Zege, E.P. Katsev, I.L. & Prikhach, A.S. 2004. Amplitude-modulated laser imager. Appl. Opt. 43: 3874–3892. Mullen, L.J. Laux, A.E. Cochenour, B. & Zege, E.P. 2006. FAMIS (Frequency Agile Modulated Imaging System) sensor for imaging in turbid water. Proc. of SPIE 6204: 62040E-1–62040E-13. Mullen, L. Cochenour, B. Rabinovich, W. Mahon, R. & Muth, J. 2009. Backscatter suppression for underwater modulating retroreflector links using polarization discrimination. Appl. Opt. 48: 328–337. Pellen, F. Intes, X. Olivard, P. Guern, Y. Cariou, J. & Lotrian, J. 2000. Determination of sea-water cut-off frequency by backscattering transfer function measurement. J. Phys. D: Appl. Phys. 33: 349–354. Ricci, R. Francucci, M. De Dominicis, L. Ferri de Collibus, M. Fornetti, G. Guarneri, M. Nuvoli, M. Paglia, E. & Bartolini, L. Techniques for effective optical noise rejection in amplitude-modulated laser optical radars for underwater three-dimensional imaging. EURASIP Journal on Advances in Signal Processing, special issue on Advances in Signal Processing for Maritime Applications. In press. Strand, P. 1995. Underwater electro-optical system for mine identification, Detection Technologies for Mines and Mine-like Targets. Proc. of SPIE 2496: 487–497.

226

LACONA8.indb 226

11/10/2010 5:37:53 PM